diff options
author | Michał Muskała <[email protected]> | 2018-08-04 23:00:55 +0200 |
---|---|---|
committer | Michał Muskała <[email protected]> | 2018-08-07 13:32:22 +0200 |
commit | c2073933a430434351ece355ba04d37fdad693ef (patch) | |
tree | 1f2fe7c76b722a3f0e391d647382ec60a1bb1c29 /lib | |
parent | fa6f32fa647a02e1735fca868b589c1f8b2ac5f8 (diff) | |
download | otp-c2073933a430434351ece355ba04d37fdad693ef.tar.gz otp-c2073933a430434351ece355ba04d37fdad693ef.tar.bz2 otp-c2073933a430434351ece355ba04d37fdad693ef.zip |
Eliminate double computation of next var
My compiler benchmarks on modules with huge functions, show the
next_free_variable_name call to be expensive. It turns out one
of the 3 calls to the function was completely ignored.
Diffstat (limited to 'lib')
-rw-r--r-- | lib/compiler/src/sys_core_fold.erl | 7 |
1 files changed, 0 insertions, 7 deletions
diff --git a/lib/compiler/src/sys_core_fold.erl b/lib/compiler/src/sys_core_fold.erl index ceb7d56221..617411ed56 100644 --- a/lib/compiler/src/sys_core_fold.erl +++ b/lib/compiler/src/sys_core_fold.erl @@ -115,13 +115,6 @@ module(#c_module{defs=Ds0}=Mod, Opts) -> {ok,Mod#c_module{defs=Ds1},get_warnings()}. function_1({#c_var{name={F,Arity}}=Name,B0}) -> - %% Find a suitable starting value for the variable counter. Note - %% that this pass assumes that new_var_name/1 returns a variable - %% name distinct from any variable used in the entire body of - %% the function. We use integers as variable names to avoid - %% filling up the atom table when compiling huge functions. - Count = cerl_trees:next_free_variable_name(B0), - put(new_var_num, Count), try %% Find a suitable starting value for the variable %% counter. Note that this pass assumes that new_var_name/1 |