Re: [jamboost] compile.c: compile_on(): Why search()?

From: Rene Rivera (grafik666_at_[hidden])
Date: 2002-11-18 12:21:18

[2002-11-18] David Abrahams wrote:

>Perhaps we should investigate this...
>
>
>From: Ingo Weinhold <bonefish_at_[hidden]>
>
>Hi,
>
>I'm currently doing a bit of profiling to optimize a rather large build
>system. And while doing so, I stumbled over the following line in
>compile_on():
>
> t->boundname = search( t->name, &t->time );
>
>Why would one do that? I somehow suspect, that it is an unneeded leftover
>from copying and pasting compile_include(). At least the unchanged comment
>for `parse->left' (preceding the function definition) seems to support
>that suspicion.
>
>For search() causes disk access (one or more timestamp()s), excessive
>use of on target rule execution can result in a significantly decreased
>build system performance.
>
>CU, Ingo

That line certainly looks like it's superfluous. Fortunately, or
unfortunately depending on point of view, it's something we won't run into.
That is unless we start using "rule-name on target-name ..." invocations.

-- grafik - Don't Assume Anything
-- rrivera_at_[hidden] - grafik_at_[hidden]
-- 102708583_at_icq - Grafik666_at_AIM - Grafik_at_[hidden]