Like -O0 , -Og completely disables a number of optimization passes so that individual options controlling them have no effect. Otherwise -Og enables all -O1 optimization flags except for those that may interfere with debugging:.

Optimize aggressively for size rather than speed. This may increase the number of instructions executed if those instructions require fewer bytes to encode. If you use multiple -O options, with or without level numbers, the last such option is the one that is effective.

Options of the form -f flag specify machine-independent flags. Most flags have both positive and negative forms; the negative form of -ffoo is -fno-foo. In the table below, only one of the forms is listed—the one you typically use. The following options control specific optimizations. They are either activated by -O options or are related to ones that are. For machines that must pop arguments after a function call, always pop the arguments as soon as each function returns.

At levels -O1 and higher, -fdefer-pop is the default; this allows the compiler to let arguments accumulate on the stack for several function calls and pop them all at once.

Perform a forward propagation pass on RTL. The pass tries to combine two instructions and checks if the result can be simplified. If loop unrolling is active, two passes are performed and the second is scheduled after loop unrolling.

This option is enabled by default at optimization levels -O1 , -O2 , -O3 , -Os. This avoids the instructions to save, set up and restore the frame pointer; on many targets it also makes an extra register available.

On some targets this flag has no effect because the standard calling sequence always uses a frame pointer, so it cannot be omitted. Several targets always omit the frame pointer in leaf functions. Optimize various standard C string functions e. This is the default when not optimizing. Single functions can be exempted from inlining by marking them with the noinline attribute. Integrate functions into their callers when their body is smaller than expected function call code so overall size of program gets smaller.

The compiler heuristically decides which functions are simple enough to be worth integrating in this way. This inlining applies to all functions, even those not declared inline. Inline also indirect calls that are discovered to be known at compile time thanks to previous inlining. This option has any effect only when inlining itself is turned on by the -finline-functions or -finline-small-functions options. Consider all functions for inlining, even if they are not declared inline.

The compiler heuristically decides which functions are worth integrating in this way. If all calls to a given function are integrated, and the function is declared static , then the function is normally not output as assembler code in its own right. Enabled at levels -O2 , -O3 , -Os. Also enabled by -fprofile-use and -fauto-profile. Consider all static functions called once for inlining into their caller even if they are not marked inline. If a call to a given function is integrated, then the function is not output as assembler code in its own right.

Enabled at levels -O1 , -O2 , -O3 and -Os , but not -Og. Doing so makes profiling significantly cheaper and usually inlining faster on programs having large chains of nested wrapper functions. Perform interprocedural scalar replacement of aggregates, removal of unused parameters and replacement of parameters passed by reference by parameters passed by value. By default, GCC limits the size of functions that can be inlined.

This flag allows coarse control of this limit. n is the size of functions that can be inlined in number of pseudo instructions. See below for a documentation of the individual parameters controlling inlining and for the defaults of these parameters. Note: there may be no value to -finline-limit that results in default behavior. In no way does it represent a count of assembly instructions and as such its exact meaning might change from one release to an another.

This is a more fine-grained version of -fkeep-inline-functions , which applies only to functions that are declared using the dllexport attribute or declspec.

See Declaring Attributes of Functions. In C, emit static functions that are declared inline into the object file, even if the function has been inlined into all of its callers. This switch does not affect functions using the extern inline extension in GNU C GCC enables this option by default. If you want to force the compiler to check if a variable is referenced, regardless of whether or not optimization is turned on, use the -fno-keep-static-consts option. Attempt to merge identical constants string constants and floating-point constants across compilation units.

This option is the default for optimized compilation if the assembler and linker support it. Use -fno-merge-constants to inhibit this behavior. This option implies -fmerge-constants. In addition to -fmerge-constants this considers e. even constant initialized arrays or initialized constant variables with integral or floating-point types.

Perform swing modulo scheduling immediately before the first scheduling pass. This pass looks at innermost loops and reorders their instructions by overlapping different iterations. Perform more aggressive SMS-based modulo scheduling with register moves allowed. By setting this flag certain anti-dependences edges are deleted, which triggers the generation of reg-moves based on the life-range analysis. This option is effective only with -fmodulo-sched enabled.

The default is -fbranch-count-reg at -O1 and higher, except for -Og. This option results in less efficient code, but some strange hacks that alter the assembler output may be confused by the optimizations performed when this option is not used. If the target supports a BSS section, GCC by default puts variables that are initialized to zero into BSS.

This can save space in the resulting code. This option turns off this behavior because some programs explicitly rely on variables going to the data section—e. Perform optimizations that check to see if a jump branches to a location where another comparison subsumed by the first is found. If so, the first branch is redirected to either the destination of the second branch or a point immediately following it, depending on whether the condition is known to be true or false.

When using a type that occupies multiple registers, such as long long on a bit system, split the registers apart and allocate them independently. This normally generates better code for those types, but may make debugging more difficult. Fully split wide types early, instead of very late.

This option has no effect unless -fsplit-wide-types is turned on. In common subexpression elimination CSE , scan through jump instructions when the target of the jump is not reached by any other path.

For example, when CSE encounters an if statement with an else clause, CSE follows the jump when the condition tested is false. This is similar to -fcse-follow-jumps , but causes CSE to follow jumps that conditionally skip over blocks.

When CSE encounters a simple if statement with no else clause, -fcse-skip-blocks causes CSE to follow the jump around the body of the if. Perform a global common subexpression elimination pass. This pass also performs global constant and copy propagation. Note: When compiling a program using computed gotos, a GCC extension, you may get better run-time performance if you disable the global common subexpression elimination pass by adding -fno-gcse to the command line. When -fgcse-lm is enabled, global common subexpression elimination attempts to move loads that are only killed by stores into themselves.

When -fgcse-sm is enabled, a store motion pass is run after global common subexpression elimination. This pass attempts to move stores out of loops.

When -fgcse-las is enabled, the global common subexpression elimination pass eliminates redundant loads that come after stores to the same memory location both partial and full redundancies. When -fgcse-after-reload is enabled, a redundant load elimination pass is performed after reload. The purpose of this pass is to clean up redundant spilling. This option tells the loop optimizer to use language constraints to derive bounds for the number of iterations of a loop.

This assumes that loop code does not invoke undefined behavior by for example causing signed integer overflows or out-of-bound array accesses. The bounds for the number of iterations of a loop are used to guide loop unrolling and peeling and loop exit test optimizations. This option is enabled by default. This option tells the compiler that variables declared in common blocks e.

Fortran may later be overridden with longer trailing arrays. This prevents certain optimizations that depend on knowing the array bounds.

Perform cross-jumping transformation. This transformation unifies equivalent code and saves code size. The resulting code may or may not perform better than without cross-jumping. Combine increments or decrements of addresses with memory accesses. This pass is always skipped on architectures that do not have instructions to support this. Enabled by default at -O1 and higher on architectures that support this. Attempt to transform conditional jumps into branch-less equivalents.

This includes use of conditional moves, min, max, set flags and abs instructions, and some tricks doable by standard arithmetics. The use of conditional execution on chips where it is available is controlled by -fif-conversion2.

Enabled at levels -O1 , -O2 , -O3 , -Os , but not with -Og. Use conditional execution where available to transform conditional jumps into branch-less equivalents. For a hierarchy with virtual bases, the base and complete variants are clones, which means two copies of the function. With this option, the base and complete variants are changed to be thunks that call a common implementation.

Assume that programs cannot safely dereference null pointers, and that no code or data element resides at address zero. This option enables simple constant folding optimizations at all optimization levels. In addition, other optimization passes in GCC use this flag to control global dataflow analyses that eliminate useless checks for null pointers; these assume that a memory access to address zero always results in a trap, so that if a pointer is checked after it has already been dereferenced, it cannot be null.

Note however that in some environments this assumption is not true. Use -fno-delete-null-pointer-checks to disable this optimization for programs that depend on that behavior.

This option is enabled by default on most targets. On Nios II ELF, it defaults to off. On AVR and MSP, this option is completely disabled. Passes that use the dataflow information are enabled independently at different optimization levels.

Attempt to convert calls to virtual functions to direct calls. This is done both within a procedure and interprocedurally as part of indirect inlining -findirect-inlining and interprocedural constant propagation -fipa-cp. Attempt to convert calls to virtual functions to speculative direct calls. Based on the analysis of the type inheritance graph, determine for a given call the set of likely targets.

If the set is small, preferably of size 1, change the call into a conditional deciding between direct and indirect calls. The speculative calls enable more optimizations, such as inlining. When they seem useless after further optimization, they are converted back into original form. Stream extra information needed for aggressive devirtualization when running the link-time optimizer in local transformation mode. This option enables more devirtualization but significantly increases the size of streamed data.

For this reason it is disabled by default. Attempt to remove redundant extension instructions. This is especially helpful for the x architecture, which implicitly zero-extends in bit registers after writing to their lower bit half.

Enabled for Alpha, AArch64 and x86 at levels -O2 , -O3 , -Os. Normally dead store elimination will take advantage of this; if your code relies on the value of the object storage persisting beyond the lifetime of the object, you can use this flag to disable this optimization. To preserve stores before the constructor starts e. Attempt to decrease register pressure through register live range shrinkage. This is helpful for fast processors with small or moderate size register sets.

Use the specified coloring algorithm for the integrated register allocator. Chaitin-Briggs coloring is not implemented for all architectures, but for those targets that do support it, it is the default because it generates better code. Use specified regions for the integrated register allocator. The region argument should be one of the following:. Use all loops as register allocation regions.

Use all loops except for loops with small register pressure as the regions. This value usually gives the best results in most cases and for most architectures, and is enabled by default when compiling with optimization for speed -O , -O2 , …. Use all functions as a single region. This typically results in the smallest code size, and is enabled by default for -Os or -O0.

Use IRA to evaluate register pressure in the code hoisting pass for decisions to hoist expressions. This option usually results in smaller code, but it can slow the compiler down. Use IRA to evaluate register pressure in loops for decisions to move loop invariants. Disable sharing of stack slots used for saving call-used hard registers living through a call. Each hard register gets a separate stack slot, and as a result function stack frames are larger.

Disable sharing of stack slots allocated for pseudo-registers. Each pseudo-register that does not get a hard register gets a separate stack slot, and as a result function stack frames are larger. Enable CFG-sensitive rematerialization in LRA. Instead of loading values of spilled pseudos, LRA tries to rematerialize recalculate values if it is profitable.

If supported for the target machine, attempt to reorder instructions to exploit instruction slots available after delayed branch instructions. Enabled at levels -O1 , -O2 , -O3 , -Os , but not at -Og. If supported for the target machine, attempt to reorder instructions to eliminate execution stalls due to required data being unavailable. This helps machines that have slow floating point or memory load instructions by allowing other instructions to be issued until the result of the load or floating-point instruction is required.

Similar to -fschedule-insns , but requests an additional pass of instruction scheduling after register allocation has been done. This is especially useful on machines with a relatively small number of registers and where memory load instructions take more than one cycle.

Disable instruction scheduling across basic blocks, which is normally enabled when scheduling before register allocation, i. with -fschedule-insns or at -O2 or higher. Disable speculative motion of non-load instructions, which is normally enabled when scheduling before register allocation, i. Enable register pressure sensitive insn scheduling before register allocation.

This only makes sense when scheduling before register allocation is enabled, i. Usage of this option can improve the generated code and decrease its size by preventing register pressure increase above the number of available hard registers and subsequent spills in register allocation. Allow speculative motion of some load instructions. This only makes sense when scheduling before register allocation, i.

Allow speculative motion of more load instructions. Define how many insns if any can be moved prematurely from the queue of stalled insns into the ready list during the second scheduling pass. Define how many insn groups cycles are examined for a dependency on a stalled insn that is a candidate for premature removal from the queue of stalled insns.

This has an effect only during the second scheduling pass, and only if -fsched-stalled-insns is used. When scheduling after register allocation, use superblock scheduling. This allows motion across basic block boundaries, resulting in faster schedules. This option is experimental, as not all machine descriptions used by GCC model the CPU closely enough to avoid unreliable results from the algorithm.

This only makes sense when scheduling after register allocation, i. with -fschedule-insns2 or at -O2 or higher. Enable the group heuristic in the scheduler. This heuristic favors the instruction that belongs to a schedule group.

This is enabled by default when scheduling is enabled, i. with -fschedule-insns or -fschedule-insns2 or at -O2 or higher. Enable the critical-path heuristic in the scheduler. This heuristic favors instructions on the critical path. Enable the speculative instruction heuristic in the scheduler.

This heuristic favors speculative instructions with greater dependency weakness. Enable the rank heuristic in the scheduler. This heuristic favors the instruction belonging to a basic block with greater size or frequency.

Enable the last-instruction heuristic in the scheduler. This heuristic favors the instruction that is less dependent on the last instruction scheduled. Enable the dependent-count heuristic in the scheduler.

This heuristic favors the instruction that has more instructions depending on it. Modulo scheduling is performed before traditional scheduling. If a loop is modulo scheduled, later scheduling passes may change its schedule. Use this option to control that behavior. Schedule instructions using selective scheduling algorithm. Selective scheduling runs instead of the first scheduler pass.

Selective scheduling runs instead of the second scheduler pass. Enable software pipelining of innermost loops during selective scheduling. This option has no effect unless one of -fselective-scheduling or -fselective-scheduling2 is turned on.

When pipelining loops during selective scheduling, also pipeline outer loops. This option has no effect unless -fsel-sched-pipelining is turned on. Some object formats, like ELF, allow interposing of symbols by the dynamic linker. This means that for symbols exported from the DSO, the compiler cannot perform interprocedural propagation, inlining and other optimizations in anticipation that the function or variable in question may change.

While this feature is useful, for example, to rewrite memory allocation functions by a debugging implementation, it is expensive in the terms of code quality. With -fno-semantic-interposition the compiler assumes that if interposition happens for functions the overwriting function will have precisely the same semantics and side effects.

Similarly if interposition happens for variables, the constructor of the variable will be the same. The flag has no effect for functions explicitly declared inline where it is never allowed for interposition to change semantics and for symbols explicitly declared weak. Emit function prologues only before parts of the function that need it, rather than at the top of the function.

This flag is enabled by default at -O and higher. Shrink-wrap separate parts of the prologue and epilogue separately, so that those parts are only executed when needed.

This option is on by default, but has no effect unless -fshrink-wrap is also turned on and the target supports this. Enable allocation of values to registers that are clobbered by function calls, by emitting extra instructions to save and restore the registers around such calls.

Such allocation is done only when it seems to result in better code. This option is always enabled by default on certain machines, usually those which have no call-preserved registers to use instead. Tracks stack adjustments pushes and pops and stack memory references and then tries to find ways to combine them. Use caller save registers for allocation if those registers are not used by any called function.

In that case it is not necessary to save and restore them around calls. This is only possible if called functions are part of same compilation unit as current function and they are compiled before it. Attempt to minimize stack usage. The compiler attempts to use less stack space, even if that makes the program slower.

This option implies setting the large-stack-frame parameter to and the large-stack-frame-growth parameter to Perform code hoisting.

Code hoisting tries to move the evaluation of expressions executed on all paths to the function exit as early as possible. This is especially useful as a code size optimization, but it often helps for code speed as well. This flag is enabled by default at -O2 and higher.

Perform partial redundancy elimination PRE on trees. This flag is enabled by default at -O2 and -O3. Make partial redundancy elimination PRE more aggressive. This flag is enabled by default at -O3. Perform forward propagation on trees. This flag is enabled by default at -O1 and higher. Perform full redundancy elimination FRE on trees. The difference between FRE and PRE is that FRE only considers expressions that are computed on all paths leading to the redundant computation.

This analysis is faster than PRE, though it exposes fewer redundancies. Perform hoisting of loads from conditional pointers on trees. This pass is enabled by default at -O1 and higher. Speculatively hoist loads from both branches of an if-then-else if the loads are from adjacent locations in the same structure and the target architecture has a conditional move instruction.

Perform copy propagation on trees. This pass eliminates unnecessary copy operations. Discover which functions are pure or constant.

Enabled by default at -O1 and higher. Discover which static variables do not escape the compilation unit. Discover read-only, write-only and non-addressable static variables. Perform interprocedural pointer analysis and interprocedural modification and reference analysis. This option can cause excessive memory and compile-time usage on large compilation units. It is not enabled by default at any optimization level. Perform interprocedural profile propagation.

The functions called only from cold functions are marked as cold. Also functions executed once such as cold , noreturn , static constructors or destructors are identified. Cold functions and loop less parts of functions executed once are then optimized for size. This optimization analyzes the side effects of functions memory locations that are modified or referenced and enables better optimization across the function call boundary.

Perform interprocedural constant propagation. This optimization analyzes the program to determine when values passed to functions are constants and then optimizes accordingly.

This optimization can substantially increase performance if the application has constants passed to functions. This flag is enabled by default at -O2 , -Os and -O3.

It is also enabled by -fprofile-use and -fauto-profile. Perform function cloning to make interprocedural constant propagation stronger. When enabled, interprocedural constant propagation performs function cloning when externally visible function can be called with constant arguments. When enabled, perform interprocedural bitwise constant propagation. This flag is enabled by default at -O2 and by -fprofile-use and -fauto-profile.

It requires that -fipa-cp is enabled. When enabled, perform interprocedural propagation of value ranges. This flag is enabled by default at -O2. Perform Identical Code Folding for functions and read-only variables. The optimization reduces code size and may disturb unwind stacks by replacing a function by equivalent one with a different name. The optimization works more effectively with link-time optimization enabled.

If a function is patched, its impacted functions should be patched too. Usually, the more IPA optimizations enabled, the larger the number of impacted functions for each function.

In order to control the number of impacted functions and more easily compute the list of impacted function, IPA optimizations can be partially enabled at two different levels. Only enable inlining and cloning optimizations, which includes inlining, cloning, interprocedural scalar replacement of aggregates and partial inlining. Only enable inlining of static functions. As a result, when patching a static function, all its callers are impacted and so need to be patched as well. When -flive-patching is specified without any value, the default value is inline-clone.

Note that -flive-patching is not supported with link-time optimization -flto. Detect paths that trigger erroneous or undefined behavior due to dereferencing a null pointer.

Isolate those paths from the main control flow and turn the statement with erroneous or undefined behavior into a trap. This flag is enabled by default at -O2 and higher and depends on -fdelete-null-pointer-checks also being enabled. This is not currently enabled, but may be enabled by -O2 in the future. Perform forward store motion on trees. Perform sparse conditional bit constant propagation on trees and propagate pointer alignment information. This pass only operates on local scalar variables and is enabled by default at -O1 and higher, except for -Og.

It requires that -ftree-ccp is enabled. Perform sparse conditional constant propagation CCP on trees. This pass only operates on local scalar variables and is enabled by default at -O1 and higher. Propagate information about uses of a value up the definition chain in order to simplify the definitions.

For example, this pass strips sign operations if the sign of a value never matters. The flag is enabled by default at -O1 and higher.

Perform pattern matching on SSA PHI nodes to optimize conditional code. This pass is enabled by default at -O1 and higher, except for -Og. Perform conversion of simple initializations in a switch to initializations from a scalar array. Look for identical code sequences.

When found, replace one with a jump to the other. This optimization is known as tail merging or cross jumping. The compilation time in this pass can be limited using max-tail-merge-comparisons parameter and max-tail-merge-iterations parameter. Perform dead code elimination DCE on trees.

Perform conditional dead code elimination DCE for calls to built-in functions that may set errno but are otherwise free of side effects. This flag is enabled by default at -O2 and higher if -Os is not also specified.

Assume that a loop with an exit will eventually take the exit and not loop indefinitely. This allows the compiler to remove loops that otherwise have no side-effects, not considering eventual endless looping as such. This also performs jump threading to reduce jumps to jumps. Perform dead store elimination DSE on trees. A dead store is a store into a memory location that is later overwritten by another store without any intervening loads. In this case the earlier store can be deleted.

Perform loop header copying on trees. This is beneficial since it increases effectiveness of code motion optimizations. It also saves one jump. It is not enabled for -Os , since it usually increases code size. Perform loop optimizations on trees. Perform loop nest optimizations. Same as -floop-nest-optimize. To use this code transformation, GCC has to be configured with --with-isl to enable the Graphite loop transformation infrastructure.

Enable the identity transformation for graphite. For every SCoP we generate the polyhedral representation and transform it back to gimple. Some minimal optimizations are also performed by the code generator isl, like index splitting and dead code elimination in loops.

Enable the isl based loop nest optimizer. This is a generic loop nest optimizer based on the Pluto optimization algorithms. It calculates a loop structure optimized for data-locality and parallelism.

This option is experimental. Use the Graphite data dependence analysis to identify loops that can be parallelized. Parallelize all the loops that can be analyzed to not contain loop carried dependences without checking that it is profitable to parallelize the loops.

While transforming the program out of the SSA representation, attempt to reduce copying by coalescing versions of different user-defined variables, instead of just compiler temporaries.

This may severely limit the ability to debug an optimized program compiled with -fno-var-tracking-assignments. In the negated form, this flag prevents SSA coalescing of user variables. This option is enabled by default if optimization is enabled, and it does very little otherwise. Attempt to transform conditional jumps in the innermost loops to branch-less equivalents. The intent is to remove control-flow from the innermost loops in order to improve the ability of the vectorization pass to handle these loops.

This is enabled by default if vectorization is enabled. Perform loop distribution. This flag can improve cache performance on big loop bodies and allow further loop optimizations, like parallelization or vectorization, to take place.

For example, the loop. Perform loop distribution of patterns that can be code generated with calls to a library. This flag is enabled by default at -O2 and higher, and by -fprofile-use and -fauto-profile.

This pass distributes the initialization loops and generates a call to memset zero. and the initialization loop is transformed into a call to memset zero. Perform loop interchange outside of graphite.

This flag can improve cache performance on loop nest and allow further loop optimizations, like vectorization, to take place. Apply unroll and jam transformations on feasible loops. In a loop nest this unrolls the outer loop by some factor and fuses the resulting multiple inner loops.

Perform loop invariant motion on trees. This pass moves only invariants that are hard to handle at RTL level function calls, operations that expand to nontrivial sequences of insns. With -funswitch-loops it also moves operands of conditions that are invariant out of the loop, so that we can use just trivial invariantness analysis in loop unswitching. The pass also includes store motion. Create a canonical counter for number of iterations in loops for which determining number of iterations requires complicated analysis.

Later optimizations then may determine the number easily. Useful especially in connection with unrolling. Perform final value replacement.

If a variable is modified in a loop in such a way that its value when exiting the loop can be determined using only its initial value and the number of loop iterations, replace uses of the final value by such a computation, provided it is sufficiently cheap.

This reduces data dependencies and may allow further simplifications. Perform induction variable optimizations strength reduction, induction variable merging and induction variable elimination on trees. Parallelize loops, i. This is only possible for loops whose iterations are independent and can be arbitrarily reordered. The optimization is only profitable on multiprocessor machines, for loops that are CPU-intensive, rather than constrained e.

by memory bandwidth. This option implies -pthread , and thus is only supported on targets that have support for -pthread. Perform function-local points-to analysis on trees.

This flag is enabled by default at -O1 and higher, except for -Og. Perform scalar replacement of aggregates.

This pass replaces structure references with scalars to prevent committing structures to memory too early. Perform merging of narrow stores to consecutive memory addresses. This pass merges contiguous stores of immediate values narrower than a word into fewer wider stores to reduce the number of instructions. This is enabled by default at -O2 and higher as well as -Os. This results in non-GIMPLE code, but gives the expanders much more complex trees to work on resulting in better RTL generation.

This is enabled by default at -O1 and higher. Perform straight-line strength reduction on trees. This recognizes related expressions involving multiplications and replaces them by less expensive calculations when possible. Perform vectorization on trees. This flag enables -ftree-loop-vectorize and -ftree-slp-vectorize if not explicitly specified.

Perform loop vectorization on trees. This flag is enabled by default at -O2 and by -ftree-vectorize , -fprofile-use , and -fauto-profile. Perform basic block vectorization on trees. Initialize automatic variables with either a pattern or with zeroes to increase the security and predictability of a program by preventing uninitialized memory disclosure and use. With this option, GCC will also initialize any padding of automatic variables that have structure or union types to zeroes.

However, the current implementation cannot initialize automatic variables that are declared between the controlling expression and the first case of a switch statement. Using -Wtrivial-auto-var-init to report all such cases. You can control this behavior for a specific variable by using the variable attribute uninitialized see Variable Attributes. Alter the cost model used for vectorization. Alter the cost model used for vectorization of loops marked with the OpenMP simd directive.

All values of model have the same meaning as described in -fvect-cost-model and by default a cost model defined with -fvect-cost-model is used. Perform Value Range Propagation on trees. This is similar to the constant propagation pass, but instead of values, ranges of values are propagated. This allows the optimizers to remove unnecessary range checks like array bound checks and null pointer checks.

This is enabled by default at -O2 and higher. Null pointer check elimination is only done if -fdelete-null-pointer-checks is enabled. Split paths leading to loop backedges.

This can improve dead code elimination and common subexpression elimination. This is enabled by default at -O3 and above. Enables expression of values of induction variables in later iterations of the unrolled loop using the value in the first iteration. This breaks long dependency chains, thus improving efficiency of the scheduling passes. A combination of -fweb and CSE is often sufficient to obtain the same effect. However, that is not reliable in cases where the loop body is more complicated than a single basic block.

It also does not work at all on some architectures due to restrictions in the CSE pass. With this option, the compiler creates multiple copies of some local variables when unrolling a loop, which can result in superior code. Inline parts of functions. Perform predictive commoning optimization, i.

This option is enabled at level -O3. If supported by the target machine, generate instructions to prefetch memory to improve the performance of loops that access large arrays. This option may generate better or worse code; results are highly dependent on the structure of loops within the source code. Do not substitute constants for known return value of formatted output functions such as sprintf , snprintf , vsprintf , and vsnprintf but not printf of fprintf. This transformation allows GCC to optimize or even eliminate branches based on the known return value of these functions called with arguments that are either constant, or whose values are known to be in a range that makes determining the exact return value possible.

For example, when -fprintf-return-value is in effect, both the branch and the body of the if statement but not the call to snprint can be optimized away when i is a bit or smaller integer because the return value is guaranteed to be at most 8. The -fprintf-return-value option relies on other optimizations and yields best results with -O2 and above. It works in tandem with the -Wformat-overflow and -Wformat-truncation options.

The -fprintf-return-value option is enabled by default. Disable any machine-specific peephole optimizations. The difference between -fno-peephole and -fno-peephole2 is in how they are implemented in the compiler; some targets use one, some use the other, a few use both.

GCC uses heuristics to guess branch probabilities if they are not provided by profiling feedback -fprofile-arcs. These heuristics are based on the control flow graph. The default is -fguess-branch-probability at levels -O , -O2 , -O3 , -Os. Reorder basic blocks in the compiled function in order to reduce number of taken branches and improve code locality. Use the specified algorithm for basic block reordering. In addition to reordering basic blocks in the compiled function, in order to reduce number of taken branches, partitions hot and cold basic blocks into separate sections of the assembly and.

o files, to improve paging and cache locality performance. When -fsplit-stack is used this option is not enabled by default to avoid linker errors , but may be enabled explicitly if using a working linker. Reorder functions in the object file in order to improve code locality.

This is implemented by using special subsections. hot for most frequently executed functions and. unlikely for unlikely executed functions. Reordering is done by the linker so object file format must support named sections and linker must place them in a reasonable way. Allow the compiler to assume the strictest aliasing rules applicable to the language being compiled.

In particular, an object of one type is assumed never to reside at the same address as an object of a different type, unless the types are almost the same. A character type may alias any other type. Even with -fstrict-aliasing , type-punning is allowed, provided the memory is accessed through the union type. So, the code above works as expected. See Structures unions enumerations and bit-fields implementation.

However, this code might not:. Similarly, access by taking the address, casting the resulting pointer and dereferencing the result has undefined behavior, even if the cast uses a union type, e. The -fstrict-aliasing option is enabled at levels -O2 , -O3 , -Os. Controls whether rules of -fstrict-aliasing are applied across function boundaries. Note that if multiple functions gets inlined into a single function the memory accesses are no longer considered to be crossing a function boundary.

The -fipa-strict-aliasing option is enabled by default and is effective only in combination with -fstrict-aliasing. Align the start of functions to the next power-of-two greater than or equal to n , skipping up to m -1 bytes.

This ensures that at least the first m bytes of the function can be fetched by the CPU without crossing an n -byte alignment boundary. If m2 is not specified, it defaults to n2. Some assemblers only support this flag when n is a power of two; in that case, it is rounded up. If n is not specified or is zero, use a machine-dependent default.

The maximum allowed n option value is If this option is enabled, the compiler tries to avoid unnecessarily overaligning functions. It attempts to instruct the assembler to align by the amount specified by -falign-functions , but not to skip more bytes than the size of the function.

Parameters of this option are analogous to the -falign-functions option. If -falign-loops or -falign-jumps are applicable and are greater than this value, then their values are used instead.

Align loops to a power-of-two boundary. If the loops are executed many times, this makes up for any execution of the dummy padding instructions. Align branch targets to a power-of-two boundary, for branch targets where the targets can only be reached by jumping. In this case, no dummy operations need be executed. Allow the compiler to perform optimizations that may introduce new data races on stores, without proving that the variable cannot be concurrently accessed by other threads.

Does not affect optimization of local data. It is safe to use this option if it is known that global data will not be accessed by multiple threads. Examples of optimizations enabled by -fallow-store-data-races include hoisting or if-conversions that may cause a value that was already in memory to be re-written with that same value. Such re-writing is safe in a single threaded context but may be unsafe in a multi-threaded context. Note that on some processors, if-conversions may be required in order to enable vectorization.

This option is left for compatibility reasons. Do not reorder top-level functions, variables, and asm statements. Output them in the same order that they appear in the input file. When this option is used, unreferenced static variables are not removed. This option is intended to support existing code that relies on a particular ordering. For new code, it is better to use attributes when possible. Additionally -fno-toplevel-reorder implies -fno-section-anchors.

This also affects any such calls implicitly generated by the compiler. Constructs webs as commonly used for register allocation purposes and assign each web individual pseudo register.

This allows the register allocation pass to operate on pseudos directly, but also strengthens several other optimization passes, such as CSE, loop optimizer and trivial dead code remover. Assume that the current compilation unit represents the whole program being compiled.

This option should not be used in combination with -flto. Instead relying on a linker plugin should provide safer and more precise information.

This option runs the standard link-time optimizer. When the object files are linked together, all the function bodies are read from these ELF sections and instantiated as if they had been part of the same translation unit. To use the link-time optimizer, -flto and optimization options should be specified at compile time and during the final link. It is recommended that you compile all the files participating in the same link with the same options and also specify those options at link time.

For example:. The first two invocations to GCC save a bytecode representation of GIMPLE into special ELF sections inside foo. o and bar. The final invocation reads the GIMPLE bytecode from foo. o , merges the two files into a single internal image, and compiles the result as usual.

Since both foo. o are merged into a single image, this causes all the interprocedural analyses and optimizations in GCC to work across the two files as if they were a single one. This means, for example, that the inliner is able to inline functions in bar. o into functions in foo. o and vice-versa. The above generates bytecode for foo. c and bar. c , merges them together into a single GIMPLE representation and optimizes them as usual to produce myprog. The important thing to keep in mind is that to enable link-time optimizations you need to use the GCC driver to perform the link step.

GCC automatically performs link-time optimization if any of the objects involved were compiled with the -flto command-line option. You can always override the automatic decision to do link-time optimization by passing -fno-lto to the link command.

To make whole program optimization effective, it is necessary to make certain whole program assumptions. The compiler needs to know what functions and variables can be accessed by libraries and runtime outside of the link-time optimized unit. When supported by the linker, the linker plugin see -fuse-linker-plugin passes information to the compiler about used and externally visible symbols. When the linker plugin is not available, -fwhole-program should be used to allow the compiler to make these assumptions, which leads to more aggressive optimization decisions.

When a file is compiled with -flto without -fuse-linker-plugin , the generated object file is larger than a regular object file because it contains GIMPLE bytecodes and the usual final code see -ffat-lto-objects. This means that object files with LTO information can be linked as normal object files; if -fno-lto is passed to the linker, no interprocedural optimizations are applied.

Note that when -fno-fat-lto-objects is enabled the compile stage is faster but you cannot perform a regular, non-LTO link on them. When producing the final binary, GCC only applies link-time optimizations to those files that contain bytecode. Therefore, you can mix and match object files and libraries with GIMPLE bytecodes and final object code. GCC automatically selects which files to optimize in LTO mode and which files to link without further processing.

Generally, options specified at link time override those specified at compile time, although in some cases GCC attempts to infer link-time options from the settings used to compile the input files. If you do not specify an optimization level option -O at link time, then GCC uses the highest optimization level used when compiling the object files.

Note that it is generally ineffective to specify an optimization level option only at link time and not at compile time, for two reasons. First, compiling without optimization suppresses compiler passes that gather information needed for effective optimization at link time.

Second, some early optimization passes can be performed only at compile time and not at link time. There are some code generation flags preserved by GCC when generating bytecodes, as they need to be used during the final link. Currently, the following options and their settings are taken from the first object file that explicitly specifies them: -fcommon , -fexceptions , -fnon-call-exceptions , -fgnu-tm and all the -m target flags. The following options -fPIC , -fpic , -fpie and -fPIE are combined based on the following scheme:.

Certain ABI-changing flags are required to match in all compilation units, and trying to override this at link time with a conflicting value is ignored. This includes options such as -freg-struct-return and -fpcc-struct-return. Other options such as -ffp-contract , -fno-strict-overflow , -fwrapv , -fno-trapv or -fno-strict-aliasing are passed through to the link stage and merged conservatively for conflicting translation units.

You can override them at link time. Diagnostic options such as -Wstringop-overflow are passed through to the link stage and their setting matches that of the compile-step at function granularity. Otherwise filenames are constructed from the SHA1 hash of the contents. Specifies a custom abbreviations file, with abbreviations one to a line.

If this option is not specified, pandoc will read the data file abbreviations from the user data directory or fall back on a system default. The only use pandoc makes of this list is in the Markdown reader.

Strings found in this list will be followed by a nonbreaking space, and the period will not produce sentence-ending space in formats like LaTeX. The strings may not contain spaces. Print diagnostic output tracing parser progress to stderr. This option is intended for use by developers in diagnosing performance issues. Produce output with an appropriate header and footer e. a standalone HTML, LaTeX, TEI, or RTF file, not a fragment.

This option is set automatically for pdf , epub , epub3 , fb2 , docx , and odt output. For native output, this option causes metadata to be included; otherwise, metadata is suppressed. Use the specified file as a custom template for the generated document. Implies --standalone. See Templates , below, for a description of template syntax.

html for HTML output. If the template is not found, pandoc will search for it in the templates subdirectory of the user data directory see --data-dir. Set the template variable KEY to the value VAL when rendering the document in standalone mode. If no VAL is specified, the key will be given the value true. Run pandoc in a sandbox, limiting IO operations in readers and writers to reading the files specified on the command line.

Note that this option does not limit IO operations by filters or in the production of PDF documents. But it does offer security against, for example, disclosure of files through the use of include directives. Anyone using pandoc on untrusted user input should use this option.

Note: some readers and writers e. If these are stored on the file system, then pandoc will not be able to find them when run in --sandbox mode and will raise an error. Print the system default template for an output FORMAT. See -t for a list of possible FORMAT s. Templates in the user data directory are ignored.

Note that some of the default templates use partials, for example styles. Print a system default data file. Files in the user data directory are ignored. The default is native. Technically, the correct term would be ppi: pixels per inch. The default is 96dpi. When images contain information about dpi internally, the encoded value is used instead of the default specified by this option.

Determine how text is wrapped in the output the source code, not the rendered version. With auto the default , pandoc will attempt to wrap lines to the column width specified by --columns default With none , pandoc will not wrap lines at all. With preserve , pandoc will attempt to preserve the wrapping from the source document that is, where there are nonsemantic newlines in the source, there will be nonsemantic newlines in the output as well.

In ipynb output, this option affects wrapping of the contents of markdown cells. Specify length of lines in characters. This affects text wrapping in the generated source code see --wrap. It also affects calculation of column widths for plain text tables see Tables below. Include an automatically generated table of contents or, in the case of latex , context , docx , odt , opendocument , rst , or ms , an instruction to create one in the output document.

Note that if you are producing a PDF via ms , the table of contents will appear at the beginning of the document, before the title. Specify the number of section levels to include in the table of contents. The default is 3 which means that level-1, 2, and 3 headings will be listed in the contents. Strip out HTML comments in the Markdown or Textile source, rather than passing them on to Markdown, Textile or HTML output as raw HTML.

Disables syntax highlighting for code blocks and inlines, even when a language attribute is given. Specifies the coloring style to be used in highlighted source code. Options are pygments the default , kate , monochrome , breezeDark , espresso , zenburn , haddock , and tango. For more information on syntax highlighting in pandoc, see Syntax highlighting , below.

See also --list-highlight-styles. Instead of a STYLE name, a JSON file with extension. theme may be supplied. This will be parsed as a KDE syntax highlighting theme and if valid used as the highlighting style. To generate the JSON version of an existing style, use --print-highlight-style. Prints a JSON version of a highlighting style, which can be modified, saved with a. theme extension, and used with --highlight-style. Instructs pandoc to load a KDE XML syntax definition file, which will be used for syntax highlighting of appropriately marked code blocks.

This can be used to add support for new languages or to use altered syntax definitions for existing languages. This option may be repeated to add multiple syntax definitions. Include contents of FILE , verbatim, at the end of the header. This can be used, for example, to include special CSS or JavaScript in HTML documents. This option can be used repeatedly to include multiple files in the header. They will be included in the order specified.

Include contents of FILE , verbatim, at the beginning of the document body e. This can be used to include navigation bars or banners in HTML documents. This option can be used repeatedly to include multiple files. List of paths to search for images and other resources. The paths should be separated by : on Linux, UNIX, and macOS systems, and by ; on Windows. If --resource-path is not specified, the default resource path is the working directory.

Note that, if --resource-path is specified, the working directory must be explicitly listed or it will not be searched. This option can be used repeatedly. Search path components that come later on the command line will be searched before those that come earlier, so --resource-path foo:bar --resource-path baz:bim is equivalent to --resource-path baz:bim:foo:bar.

Set the request header NAME to the value VAL when making HTTP requests for example, when a URL is given on the command line, or when resources used in a document must be downloaded. Disable the certificate verification to allow access to unsecure HTTP resources for example when the certificate is no longer valid or self signed. Deprecated synonym for --embed-resources --standalone. Produce a standalone HTML file with no external dependencies, using data: URIs to incorporate the contents of linked scripts, stylesheets, images, and videos.

Scripts, images, and stylesheets at absolute URLs will be downloaded; those at relative URLs will be sought relative to the working directory if the first source file is local or relative to the base URL if the first source file is remote.

Limitation: resources that are loaded dynamically through JavaScript cannot be incorporated; as a result, some advanced features e. js slide show. This option only has an effect if the smart extension is enabled for the input format used. Use only ASCII characters in output. Currently supported for XML and HTML formats which use entities instead of UTF-8 when this option is selected , CommonMark, gfm, and Markdown which use entities , roff ms which use hexadecimal escapes , and to a limited degree LaTeX which uses standard commands for accented characters when possible.

roff man output uses ASCII by default. Use reference-style links, rather than inline links, in writing Markdown or reStructuredText. By default inline links are used. The placement of link references is affected by the --reference-location option. Specify whether footnotes and references, if reference-links is set are placed at the end of the current top-level block, the current section, or the document.

The default is document. Currently this option only affects the markdown , muse , html , epub , slidy , s5 , slideous , dzslides , and revealjs writers. Specify whether to use ATX-style -prefixed or Setext-style underlined headings for level 1 and 2 headings in Markdown output. The default is atx. This option also affects Markdown cells in ipynb output. Treat top-level headings as the given division type in LaTeX, ConTeXt, DocBook, and TEI output.

The hierarchy order is part, chapter, then section; all headings are shifted such that the top-level heading becomes the specified type. The default behavior is to determine the best division type via heuristics: unless other conditions apply, section is chosen. When the documentclass variable is set to report , book , or memoir unless the article option is specified , chapter is implied as the setting for this option.

Number section headings in LaTeX, ConTeXt, HTML, Docx, ms, or EPUB output. By default, sections are not numbered. Sections with class unnumbered will never be numbered, even if --number-sections is specified. Offset for section headings in HTML output ignored in other output formats. The first number is added to the section number for top-level headings, the second for second-level headings, and so on.

Offsets are 0 by default. Implies --number-sections. Use the listings package for LaTeX code blocks. The package does not support multi-byte encoding for source code. To handle UTF-8 you would need to use a custom template.

This issue is fully documented here: Encoding issue with the listings package. Make list items in slide shows display incrementally one by one. The default is for lists to be displayed all at once. Specifies that headings with the specified level create slides for beamer , s5 , slidy , slideous , dzslides.

Headings above this level in the hierarchy are used to divide the slide show into sections; headings below this level create subheads within a slide. Valid values are If a slide level of 0 is specified, slides will not be split automatically on headings, and horizontal rules must be used to indicate slide boundaries.

If a slide level is not specified explicitly, the slide level will be set automatically based on the contents of the document; see Structuring the slide show. See Heading identifiers , below. Specify a method for obfuscating mailto: links in HTML documents. none leaves mailto: links as they are. javascript obfuscates them using JavaScript. references obfuscates them by printing their letters as decimal or hexadecimal character references. The default is none.

Specify a prefix to be added to all identifiers and internal links in HTML and DocBook output, and to footnote numbers in Markdown and Haddock output. This is useful for preventing duplicate identifiers when generating fragments to be included in other pages. Specify STRING as a prefix at the beginning of the title that appears in the HTML header but not in the title as it appears at the beginning of the HTML body. Link to a CSS style sheet. A stylesheet is required for generating EPUB.

If none is provided using this option or the css or stylesheet metadata fields , pandoc will look for a file epub. css in the user data directory see --data-dir. If it is not found there, sensible defaults will be used. For best results, the reference docx should be a modified version of a docx file produced using pandoc. The contents of the reference docx are ignored, but its stylesheets and document properties including margins, page size, header, and footer are used in the new docx.

If no reference docx is specified on the command line, pandoc will look for a file reference. docx in the user data directory see --data-dir. If this is not found either, sensible defaults will be used. To produce a custom reference. docx , first get a copy of the default reference. docx : pandoc -o custom-reference. docx --print-default-data-file reference. Then open custom-reference. docx in Word, modify the styles as you wish, and save the file. For best results, do not make changes to this file other than modifying the styles used by pandoc:.

For best results, the reference ODT should be a modified version of an ODT produced using pandoc. The contents of the reference ODT are ignored, but its stylesheets are used in the new ODT.

If no reference ODT is specified on the command line, pandoc will look for a file reference. odt in the user data directory see --data-dir. odt , first get a copy of the default reference. odt : pandoc -o custom-reference. odt --print-default-data-file reference.

odt in LibreOffice, modify the styles as you wish, and save the file. Templates included with Microsoft PowerPoint either with. pptx or. potx extension are known to work, as are most templates derived from these. The specific requirement is that the template should contain layouts with the following names as seen within PowerPoint :. For each name, the first layout found with that name will be used. If no layout is found with one of the names, pandoc will output a warning and use the layout with that name from the default reference doc instead.

How these layouts are used is described in PowerPoint layout choice. All templates included with a recent version of MS PowerPoint will fit these criteria. You can click on Layout under the Home menu to check. You can also modify the default reference.

pptx : first run pandoc -o custom-reference. pptx --print-default-data-file reference. pptx , and then modify custom-reference. pptx in MS PowerPoint pandoc will use the layouts with the names listed above. Use the specified image as the EPUB cover. It is recommended that the image be less than px in width and height. Note that in a Markdown source document you can also specify cover-image in a YAML metadata block see EPUB Metadata , below.

Look in the specified XML file for metadata for the EPUB. The file should contain a series of Dublin Core elements. For example:. Any of these may be overridden by elements in the metadata file.

Note: if the source document is Markdown, a YAML metadata block in the document can be used instead. See below under EPUB Metadata. Embed the specified font in the EPUB. This option can be repeated to embed multiple fonts. However, if you use wildcards on the command line, be sure to escape them or put the whole filename in single quotes, to prevent them from being interpreted by the shell.

To use the embedded fonts, you will need to add declarations like the following to your CSS see --css :. The default is to split into chapters at level-1 headings. This option only affects the internal composition of the EPUB, not the way chapters and sections are displayed to users. Some readers may be slow if the chapter files are too large, so for large documents with few level-1 headings, one might want to use a chapter level of 2 or 3.

Specify the subdirectory in the OCF container that is to hold the EPUB-specific contents. The default is EPUB. To put the EPUB contents in the top level, use an empty string.

Determines how ipynb output cells are treated. all means that all of the data formats included in the original are preserved. none means that the contents of data cells are omitted. best causes pandoc to try to pick the richest data block in each output cell that is compatible with the output format.

The default is best. Use the specified engine when producing PDF output. Valid values are pdflatex , lualatex , xelatex , latexmk , tectonic , wkhtmltopdf , weasyprint , pagedjs-cli , prince , context , and pdfroff. If the engine is not in your PATH, the full path of the engine may be specified here.

Use the given string as a command-line argument to the pdf-engine. Note that no check for duplicate options is done. Process the citations in the file, replacing them with rendered citations and adding a bibliography. Citation processing will not take place unless bibliographic data is supplied, either through an external file specified using the --bibliography option or the bibliography field in metadata, or via a references section in metadata containing a list of citations in CSL YAML format with Markdown formatting.

The style is controlled by a CSL stylesheet specified using the --csl option or the csl field in metadata. If no stylesheet is specified, the chicago-author-date style will be used by default. The citation processing transformation may be applied before or after filters or Lua filters see --filter , --lua-filter : these transformations are applied in the order they appear on the command line. For more information, see the section on Citations.

If you supply this argument multiple times, each FILE will be added to bibliography. If FILE is a URL, it will be fetched via HTTP. If FILE is not found relative to the working directory, it will be sought in the resource path see --resource-path.

If FILE is not found relative to the working directory, it will be sought in the resource path see --resource-path and finally in the csl subdirectory of the pandoc user data directory.

Use natbib for citations in LaTeX output. This option is not for use with the --citeproc option or with PDF output. It is intended for use in producing a LaTeX file that can be processed with bibtex. Use biblatex for citations in LaTeX output. It is intended for use in producing a LaTeX file that can be processed with bibtex or biber. The default is to render TeX math as far as possible using Unicode characters. However, this gives acceptable results only for basic math, usually you will want to use --mathjax or another of the following options.

Use MathJax to display embedded TeX math in HTML output. Then the MathJax JavaScript will render it. The URL should point to the MathJax.

js load script. If a URL is not provided, a link to the Cloudflare CDN will be inserted. Convert TeX math to MathML in epub3 , docbook4 , docbook5 , jats , html4 and html5. This is the default in odt output.

Note that currently only Firefox and Safari and select e-book readers natively support MathML. The formula will be URL-encoded and concatenated with the URL provided.

Use KaTeX to display embedded TeX math in HTML output. The URL is the base URL for the KaTeX library. That directory should contain a katex. js and a katex. css file. If a URL is not provided, a link to the KaTeX CDN will be inserted. The resulting HTML can then be processed by GladTeX to produce SVG images of the typeset formulas and an HTML file with these images embedded. Print information about command-line arguments to stdout , then exit. This option is intended primarily for use in wrapper scripts.

The first line of output contains the name of the output file specified with the -o option, or - for stdout if no output file was specified. The remaining lines contain the command-line arguments, one per line, in the order they appear. These do not include regular pandoc options and their arguments, but do include any options appearing after a -- separator at the end of the line. Ignore command-line arguments for use in wrapper scripts. Regular pandoc options are not ignored.

If pandoc completes successfully, it will return exit code 0. Nonzero exit codes have the following meanings:. The --defaults option may be used to specify a package of options, in the form of a YAML file. Fields that are omitted will just have their regular default values. So a defaults file can be as simple as one line:. In fields that expect a file path or list of file paths , the following syntax may be used to interpolate environment variables:. This allows you to refer to resources contained in that directory:.

This environment variable interpolation syntax only works in fields that expect file paths. Defaults files can be placed in the defaults subdirectory of the user data directory and used from any directory.

For example, one could create a file specifying defaults for writing letters, save it as letter. yaml in the defaults subdirectory of the user data directory, and then invoke these defaults from any directory using pandoc --defaults letter or pandoc -dletter.

Note that, where command-line arguments may be repeated --metadata-file , --css , --include-in-header , --include-before-body , --include-after-body , --variable , --metadata , --syntax-definition , the values specified on the command line will combine with values specified in the defaults file, rather than replacing them. The value of input-files may be left empty to indicate input from stdin, and it can be an empty sequence [] for no input.

Options specified in a defaults file itself always have priority over those in another file included with a defaults: entry. verbosity can have the values ERROR , WARNING , or INFO. Metadata values specified in a defaults file are parsed as literal string text, not Markdown. Filters will be assumed to be Lua filters if they have the. lua extension, and JSON filters otherwise. But the filter type can also be specified explicitly, as shown. Filters are run in the order specified.

To include the built-in citeproc filter, use either citeproc or {type: citeproc}. cite-method can be citeproc , natbib , or biblatex. This only affects LaTeX output. If you need control over when the citeproc processing is done relative to other filters, you should instead use citeproc in the list of filters see above. In addition to the values listed above, method can have the value plain.

If the command line option accepts a URL argument, an url: field can be added to html-math-method:. To see the default template that is used, just type. where FORMAT is the name of the output format. A custom template can be specified using the --template option. Templates contain variables , which allow for the inclusion of arbitrary information at any point in the file.

In addition, some variables are given default values by pandoc. If you use custom templates, you may need to revise them as pandoc changes. We recommend tracking the changes in the default templates, and modifying your custom templates accordingly.

An easy way to do this is to fork the pandoc-templates repository and merge in changes after each pandoc release. The styles may also be mixed in the same template, but the opening and closing delimiter must match in each case.

The opening delimiter may be followed by one or more spaces or tabs, which will be ignored. The closing delimiter may be followed by one or more spaces or tabs, which will be ignored. A slot for an interpolated variable is a variable name surrounded by matched delimiters. The keywords it , if , else , endif , for , sep , and endfor may not be used as variable names.

Variable names with periods are used to get at structured variable values. So, for example, employee. salary will return the value of the salary field of the object that is the value of the employee field. A conditional begins with if variable enclosed in matched delimiters and ends with endif enclosed in matched delimiters. It may optionally contain an else enclosed in matched delimiters.

The if section is used if variable has a non-empty value, otherwise the else section is used if present. The keyword elseif may be used to simplify complex nested conditionals:. A for loop begins with for variable enclosed in matched delimiters and ends with endfor enclosed in matched delimiters. You may optionally specify a separator between consecutive values using sep enclosed in matched delimiters.

The material between sep and the endfor is the separator. Instead of using variable inside the loop, the special anaphoric keyword it may be used. Partials subtemplates stored in different files may be included by using the name of the partial, followed by , for example:. Partials will be sought in the directory containing the main template.

The file name will be assumed to have the same extension as the main template if it lacks an extension. When calling the partial, the full name including file extension can also be used:. If a partial is not found in the directory of the template and the template path is given as a relative path, it will also be sought in the templates subdirectory of the user data directory.

If articles is an array, this will iterate over its values, applying the partial bibentry to each one. So the second example above is equivalent to. Note that the anaphoric keyword it must be used when iterating over partials. In the above examples, the bibentry partial should contain it.

title and so on instead of articles. A separator between values of an array may be specified in square brackets, immediately after the variable name or partial:. The separator in this case is literal and unlike with sep in an explicit for loop cannot contain interpolated variables or other template directives. In this example, if item. description has multiple lines, they will all be indented to line up with the first line:.

Normally, spaces in the template itself as opposed to values of the interpolated variables are not breakable, but they can be made breakable in part of the template by using the ~ keyword ended with another ~. A pipe transforms the value of a variable or partial. pairs : Converts a map or array to an array of maps, each with key and value fields.

If the original value was an array, the key will be the array index, starting with 1. length : Returns the length of the value: number of characters for a textual value, number of elements for a map or array.

reverse : Reverses a textual value or array, and has no effect on other values. first : Returns the first value of an array, if applied to a non-empty array; otherwise returns the original value. last : Returns the last value of an array, if applied to a non-empty array; otherwise returns the original value.

rest : Returns all but the first value of an array, if applied to a non-empty array; otherwise returns the original value.

allbutlast : Returns all but the last value of an array, if applied to a non-empty array; otherwise returns the original value. chomp : Removes trailing newlines and breakable space. nowrap : Disables line wrapping on breakable spaces. alpha : Converts textual values that can be read as an integer into lowercase alphabetic characters a..

z mod This can be used to get lettered enumeration from array indices. To get uppercase letters, chain with uppercase. roman : Converts textual values that can be read as an integer into lowercase roman numerals.

To get uppercase roman, chain with uppercase. left n "leftborder" "rightborder" : Renders a textual value in a block of width n , aligned to the left, with an optional left and right border. Has no effect on other values. This can be used to align material in tables. Widths are positive integers indicating the number of characters.

right n "leftborder" "rightborder" : Renders a textual value in a block of width n , aligned to the right, and has no effect on other values. center n "leftborder" "rightborder" : Renders a textual value in a block of width n , aligned to the center, and has no effect on other values.

allow identification of basic aspects of the document. Included in PDF metadata through LaTeX and ConTeXt. These can be set through a pandoc title block , which allows for multiple authors, or through a YAML metadata block :.

Note that if you just want to set PDF or HTML metadata, without including a title block in the document itself, you can set the title-meta , author-meta , and date-meta variables. By default these are set automatically, based on title , author , and date. The page title in HTML is set by pagetitle , which is equal to title by default. Additionally, any root-level string metadata, not included in ODT, docx or pptx metadata is added as a custom property.

The following YAML metadata block for instance:. will include title , author and description as standard document properties and subtitle as a custom property when converting to docx, ODT or pptx.

identifies the main language of the document using IETF language tags following the BCP 47 standard , such as en or en-GB. The Language subtag lookup tool can look up or verify these tags. This affects most formats, and controls hyphenation in PDF output when using LaTeX through babel and polyglossia or ConTeXt. Use native pandoc Divs and Spans with the lang attribute to switch the language:. the base script direction, either rtl right-to-left or ltr left-to-right. For bidirectional documents, native pandoc span s and div s with the dir attribute value rtl or ltr can be used to override the base direction in some output formats.

This may not always be necessary if the final renderer e. the browser, when generating HTML supports the Unicode Bidirectional Algorithm. To override or extend some CSS for just one document, include for example:. These affect HTML output when producing slide shows with pandoc. All reveal. js configuration options are available as variables.

To turn off boolean flags that default to true in reveal. js, use 0. These variables change the appearance of PDF slides using beamer. These variables control the visual aspects of a slide show that are not easily controlled via templates. Pandoc uses these variables when creating a PDF with a LaTeX engine.

Instead of using this option, KOMA-Script can adjust headings more extensively:. option for document class, e. oneside ; repeat for multiple options:. option for geometry package, e. option for hyperref package, e. options for package used as fontfamily ; repeat for multiple options. For example, to use the Libertine font with proportional lowercase old-style figures through the libertinus package:.

options to use with mainfont , sansfont , monofont , mathfont , CJKmainfont in xelatex and lualatex. Allow for any choices available through fontspec ; repeat for multiple options. For example, to use the TeX Gyre version of Palatino with lowercase figures:. These variables function when using BibLaTeX for citation rendering. Pandoc uses these variables when creating a PDF with ConTeXt. Pandoc uses these variables when creating a PDF with wkhtmltopdf.

The --css option also affects the output. Pandoc sets these variables automatically in response to options or document contents; users can also modify them. These vary depending on the output format, and include the following:. source and destination filenames, as given on the command line. sourcefile can also be a list if input comes from multiple files, or empty if input is from stdin. You can use the following snippet in your template to distinguish them:. Similarly, outputfile can be - if output goes to the terminal.

If you need absolute paths, use e. The behavior of some of the readers and writers can be adjusted by enabling or disabling various extensions.

The markdown reader and writer make by far the most use of extensions. In the following, extensions that also work for other formats are covered.

Note that markdown extensions added to the ipynb format affect Markdown cells in Jupyter notebooks as do command-line options like --atx-headers. Interpret straight quotes as curly quotes, as em-dashes, -- as en-dashes, and as ellipses.

Note: If you are writing Markdown, then the smart extension has the reverse effect: what would have been curly quotes comes out straight. If smart is disabled, then in reading LaTeX pandoc will parse these characters literally. In writing LaTeX, enabling smart tells pandoc to use the ligatures when possible; if smart is disabled pandoc will use unicode quotation mark and dash characters.

A heading without an explicitly specified identifier will be automatically assigned a unique identifier based on the heading text. These rules should, in most cases, allow one to determine the identifier from the heading text. The exception is when several headings have the same text; in this case, the first will get an identifier as described above; the second will get the same identifier with -1 appended; the third with -2 ; and so on.

These identifiers are used to provide link targets in the table of contents generated by the --toc --table-of-contents option. They also make it easy to provide links from one section of a document to another. A link to this section, for example, might look like this:. Note, however, that this method of providing links to sections works only in HTML, LaTeX, and ConTeXt formats. This allows entire sections to be manipulated using JavaScript or treated differently in CSS.

Accents are stripped off of accented Latin letters, and non-Latin letters are omitted. Emojis are replaced by their names. However, they can also be used with HTML input. This is handy for reading web pages formatted using MathJax, for example. By default, this is disabled for HTML input. This means that.

In Markdown output, code blocks with classes haskell and literate will be rendered using bird tracks, and block quotations will be indented one space, so they will not be treated as Haskell code. In restructured text output, code blocks with class haskell will be rendered using bird tracks. In LaTeX input, text in code environments will be parsed as Haskell code. In LaTeX output, code blocks with class haskell will be rendered inside code environments. In HTML output, code blocks with class haskell will be rendered with class literatehaskell and bird tracks.

reads literate Haskell source formatted with Markdown conventions and writes ordinary HTML without bird tracks. writes HTML with the Haskell code in bird tracks, so it can be copied and pasted as literate Haskell source. Note that GHC expects the bird tracks in the first column, so indented literate code blocks e.

inside an itemized environment will not be picked up by the Haskell compiler. Links to headings, figures and tables inside the document are substituted with cross-references that will use the name or caption of the referenced item.

The original link text is replaced once the generated document is refreshed. Text in cross-references is only made consistent with the referenced item once the document has been refreshed. Links to headings, figures and tables inside the document are substituted with cross-references that will use the number of the referenced item. The original link text is discarded. Numbers in cross-references are only visible in the final document once it has been refreshed. When converting from docx, read all docx styles as divs for paragraph styles and spans for character styles regardless of whether pandoc understands the meaning of these styles.

This can be used with docx custom styles. Disabled by default. In the muse input format, this enables Text::Amuse extensions to Emacs Muse markup. In the ipynb input format, this causes Markdown cells to be included as raw Markdown blocks allowing lossless round-tripping rather than being parsed. Use this only when you are targeting ipynb or a markdown-based output format. When the citations extension is enabled in org , org-cite and org-ref style citations will be parsed as native pandoc citations.

When citations is enabled in docx , citations inserted by Zotero or Mendeley or EndNote plugins will be parsed as native pandoc citations.

Otherwise, the formatted citations generated by the bibliographic software will be parsed as regular text. As in Org Mode, enabling this extension allows lowercase and uppercase alphabetical markers for ordered lists to be parsed in addition to arabic ones.

These elements are not influenced by CSL styles, but all information on the item is included in tags. In the context output format this enables the use of Natural Tables TABLE instead of the default Extreme Tables xtables. Natural tables allow more fine-grained global customization but come at a performance penalty compared to extreme tables. This document explains the syntax, noting differences from original Markdown.

Extensions can be enabled or disabled to specify the behavior more granularly. They are described in the following. See also Extensions above, for extensions that work also on other formats. Whereas Markdown was originally designed with HTML generation in mind, pandoc is designed for multiple output formats. Thus, while pandoc allows the embedding of raw HTML, it discourages it, and provides other, non-HTMLish ways of representing important document elements like definition lists, tables, mathematics, and footnotes.

A paragraph is one or more lines of text followed by one or more blank lines. Newlines are treated as spaces, so you can reflow your paragraphs as you like.

If you need a hard line break, put two or more spaces at the end of a line. A backslash followed by a newline is also a hard line break. Note: in multiline and grid table cells, this is the only way to create a hard line break, since trailing spaces in the cells are ignored.

The heading text can contain inline formatting, such as emphasis see Inline formatting , below. An ATX-style heading consists of one to six signs and a line of text, optionally followed by any number of signs.

The number of signs at the beginning of the line is the heading level:. Original Markdown syntax does not require a blank line before a heading. Pandoc does require this except, of course, at the beginning of the document. The reason for the requirement is that it is all too easy for a to end up at the beginning of a line by accident perhaps through line wrapping.

Consider, for example:. Many Markdown implementations do not require a space between the opening s of an ATX heading and the heading text, so that 5 bolt and hashtag count as headings. With this extension, pandoc does require the space. Headings can be assigned attributes using this syntax at the end of the line containing the heading text:. Thus, for example, the following headings will all be assigned the identifier foo :. This syntax is compatible with PHP Markdown Extra.

Identifiers are used for labels and link anchors in the LaTeX, ConTeXt, Textile, Jira markup, and AsciiDoc writers. Headings with the class unnumbered will not be numbered, even if --number-sections is specified. A single hyphen - in an attribute context is equivalent to. unnumbered , and preferable in non-English documents.

If the unlisted class is present in addition to unnumbered , the heading will not be included in a table of contents. Currently this feature is only implemented for certain formats: those based on LaTeX and HTML, PowerPoint, and RTF. Pandoc behaves as if reference links have been defined for each heading. So, to link to a heading. If there are multiple headings with identical text, the corresponding reference will link to the first one only, and you will need to use explicit links to link to the others, as described above.

Explicit link reference definitions always take priority over implicit heading references. So, in the following example, the link will point to bar , not to foo :. Markdown uses email conventions for quoting blocks of text. Among the block elements that can be contained in a block quote are other block quotes. That is, block quotes can be nested:. Original Markdown syntax does not require a blank line before a block quote.

A block of text indented four spaces or one tab is treated as verbatim text: that is, special characters do not trigger special formatting, and all spaces and line breaks are preserved. For example,. The initial four space or one tab indentation is not considered part of the verbatim text, and is removed in the output. In addition to standard indented code blocks, pandoc supports fenced code blocks. These begin with a row of three or more tildes ~ and end with a row of tildes that must be at least as long as the starting row.

Everything between these lines is treated as code. No indentation is necessary:. Like regular code blocks, fenced code blocks must be separated from surrounding text by blank lines. If the code itself contains a row of tildes or backticks, just use a longer row of tildes or backticks at the start and end:. Here mycode is an identifier, haskell and numberLines are classes, and startFrom is an attribute with value Some output formats can use this information to do syntax highlighting.

Currently, the only output formats that use this information are HTML, LaTeX, Docx, Ms, and PowerPoint. If highlighting is supported for your output format and language, then the code block above will appear highlighted, with numbered lines.

To see which languages are supported, type pandoc --list-highlight-languages. Otherwise, the code block above will appear as follows:. The numberLines or number-lines class will cause the lines of the code block to be numbered, starting with 1 or the value of the startFrom attribute. The lineAnchors or line-anchors class will cause the lines to be clickable anchors in HTML output.

To prevent all highlighting, use the --no-highlight flag. To set the highlighting style, use --highlight-style. For more information on highlighting, see Syntax highlighting , below. A line block is a sequence of lines beginning with a vertical bar followed by a space. The division into lines will be preserved in the output, as will any leading spaces; otherwise, the lines will be formatted as Markdown.

This is useful for verse and addresses:. The lines can be hard-wrapped if needed, but the continuation line must begin with a space. Inline formatting such as emphasis is allowed in the content, but not block-level formatting such as block quotes or lists. This syntax is borrowed from reStructuredText. A bullet list is a list of bulleted list items.

Here is a simple example:. The bullets need not be flush with the left margin; they may be indented one, two, or three spaces. The bullet must be followed by whitespace. A list item may contain multiple paragraphs and other block-level content. However, subsequent paragraphs must be preceded by a blank line and indented to line up with the first non-space content after the list marker.

Exception: if the list marker is followed by an indented code block, which must begin 5 spaces after the list marker, then subsequent paragraphs must begin two columns after the last character of the list marker:. List items may include other lists. In this case the preceding blank line is optional. The nested list must be indented to line up with the first non-space character after the list marker of the containing list item. However, if there are multiple paragraphs or other blocks in a list item, the first line of each must be indented.

Ordered lists work just like bulleted lists, except that the items begin with enumerators rather than bullets. In original Markdown, enumerators are decimal numbers followed by a period and a space. The numbers themselves are ignored, so there is no difference between this list:. Unlike original Markdown, pandoc allows ordered list items to be marked with uppercase and lowercase letters and roman numerals, in addition to Arabic numerals.

List markers may be enclosed in parentheses or followed by a single right-parenthesis or period.

Work fast with our official CLI. Learn more. Please sign in to use Codespaces. If nothing happens, download GitHub Desktop and try again. If nothing happens, download Xcode and try again. There was a problem preparing your codespace, please try again. First carefully read the installation instructions for your OS.

We recommend you use the supplied install. py - the "full" installation guide is for rare, advanced use cases and most users should use install. If the server isn't starting and you're getting a "YouCompleteMe unavailable" error, check the Troubleshooting guide. Next check the User Guide section on the semantic completer that you are using.

Finally, check the FAQ. If, after reading the installation and user guides, and checking the FAQ, you're still having trouble, check the contacts section below for how to get in touch. Please do NOT go to vim on Freenode for support.

Please contact the YouCompleteMe maintainers directly using the contact details below. YouCompleteMe is a fast, as-you-type, fuzzy-search code completion, comprehension and refactoring engine for Vim.

It has several completion engines built in and supports any protocol-compliant Language Server, so can work with practically any language.

YouCompleteMe contains:. First, realize that no keyboard shortcuts had to be pressed to get the list of completion candidates at any point in the demo. The user just types and the suggestions pop up by themselves.

When the user sees a useful completion string being offered, they press the TAB key to accept it. This inserts the completion string.

Repeated presses of the TAB key cycle through the offered completions. If the offered completions are not relevant enough, the user can continue typing to further filter out unwanted completions.

A critical thing to notice is that the completion filtering is NOT based on the input being a string prefix of the completion but that works too. The input needs to be a subsequence match of a completion. This is a fancy way of saying that any input characters need to be present in a completion string in the order in which they appear in the input. So abc is a subsequence of xaybgc , but not of xbyxaxxc. After the filter, a complicated sorting system ranks the completion strings so that the most relevant ones rise to the top of the menu so you usually need to press TAB just once.

All of the above works with any programming language because of the identifier-based completion engine. It collects all of the identifiers in the current file and other files you visit and your tags files and searches them when you type identifiers are put into per-filetype groups.

The demo also shows the semantic engine in use. When the user presses. The last thing that you can see in the demo is YCM's diagnostic display features the little red X that shows up in the left gutter; inspired by Syntastic if you are editing a C-family file. As the completer engine compiles your file and detects warnings or errors, they will be presented in various ways.

You don't need to save your file or press any keyboard shortcut to trigger this, it "just happens" in the background. YCM might be the only vim completion engine with the correct Unicode support. Though we do assume UTF-8 everywhere. YCM also provides semantic IDE-like features in a number of languages, including:. And here's some documentation being shown in a hover popup, automatically and manually:.

Features vary by file type, so make sure to check out the file type feature summary and the full list of completer subcommands to find out what's available for your favourite languages.

You'll also find that YCM has filepath completers try typing. Our policy is to support the Vim version that's in the latest LTS of Ubuntu. That's currently Ubuntu Vim must have a working Python 3. You can check with :py3 import sys; print sys. For Neovim users, our policy is to require the latest released version. Currently, Neovim 0.

Please note that some features are not available in Neovim, and Neovim is not officially supported. This requires a version bump of the minimum supported compilers. The new requirements are:. YCM requires CMake 3. If your CMake is too old, you may be able to simply pip install --user cmake to get a really new version. When enabling language support for a particular language, there may be runtime requirements, such as needing a very recent Java Development Kit for Java support.

In general, YCM is not in control of the required versions for the downstream compilers, though we do our best to signal where we know them. Install mono from Mono Project NOTE: on Intel Macs you can also brew install mono.

On arm Macs, you may require Rosetta. Pre-installed macOS system Vim does not support Python 3. So you need to install either a Vim that supports Python 3 OR MacVim with Homebrew :. For using an arbitrary LSP server, check the relevant section. These instructions using install.

py are the quickest way to install YouCompleteMe, however they may not work for everyone. If the following instructions don't work for you, check out the full installation guide.

A supported Vim version with Python 3 is required. MacVim is a good option, even if you only use the terminal. YCM won't work with the pre-installed Vim from Apple as its Python support is broken. If you don't already use a Vim that supports Python 3 or MacVim , install it with Homebrew.

Install CMake as well:. Install YouCompleteMe with Vundle. Remember: YCM is a plugin with a compiled component. You should then rerun the install process. NOTE: If you want C-family completion, you MUST have the latest Xcode installed along with the latest Command Line Tools they are installed automatically when you run clang for the first time, or manually by running xcode-select --install.

Compiling YCM with semantic support for C-family languages through clangd :. To simply compile with everything enabled, there's a --all flag.

So, to install with all language features, ensure xbuild , go , node and npm tools are installed and in your PATH , then simply run:. That's it. You're done. Refer to the User Guide section on how to use YCM. Don't forget that if you want the C-family semantic completion engine to work, you will need to provide the compilation flags for your project to YCM.

It's all in the User Guide. YCM comes with sane defaults for its options, but you still may want to take a look at what's available for configuration. There are a few interesting options that are conservatively turned off by default that you may want to turn on.

Make sure you have a supported version of Vim with Python 3 support, and a supported compiler. The latest LTS of Ubuntu is the minimum platform for simple installation. For earlier releases or other distributions, you may have to do some work to acquire the dependencies. If your vim version is too old, you may need to compile Vim from source don't worry, it's easy.

Important: we assume that you are using the cmd. exe command prompt and that you know how to add an executable to the PATH environment variable. Make sure you have a supported Vim version with Python 3 support. You can check the version and which Python is supported by typing :version inside Vim. Take note of the Vim architecture, i. It will be important when choosing the Python installer. We recommend using a bit client. Daily updated installers of bit and bit Vim with Python 3 support are available.

Add the following line to your vimrc if not already present. This option is required by YCM. Note that it does not prevent you from editing a file in another encoding than UTF So, to install with all language features, ensure msbuild , go , node and npm tools are installed and in your PATH , then simply run:. YCM officially supports MSVC 15 , MSVC 16 Visual Studio and MSVC 17 Visual Studio 17

YCM supports completion in buffers with no filetype set, but this must be explicitly whitelisted. Use the specified algorithm for basic block reordering. This option is reserved , meaning that while signature help support remains experimental, its values and meaning may change and it may be removed in a future version. Field names must not be interpretable as YAML numbers or boolean values so, for example, yes , True , and 15 cannot be used as field names. The strip length can be changed using the loop-block-tile-size parameter.

Categories:

• Binary Options online
• Forex
• Forex Brokers
• Binary Options
• Forex Bitcoin