automake-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Patch: speed up compilations


From: Alexandre Duret-Lutz
Subject: Re: Patch: speed up compilations
Date: Sat, 06 Jul 2002 13:02:31 +0200
User-agent: Gnus/5.090007 (Oort Gnus v0.07) Emacs/21.2 (i386-debian-linux-gnu)

>>> "Tom" == Tom Tromey <address@hidden> writes:

 Tom> This patch changes automake so that dependency-tracking compilations
 Tom> in gcc3 mode are inlined.  This gives us a speedup in this common and
 Tom> imho important case.

That's great.

 Tom> Unfortunately this patch isn't perfect :-(.  It regresses cxx2.test,
 Tom> at least.

 Tom> Alexandre, what do you think is the best way to fix the resulting
 Tom> failure?  

`cvs up' :) A side effect of the monstro-patch I've just
committed is that `automake -Werror' no longer abort after the
first error.

 Tom> It looks like this:

 Tom> /home/tromey/gnu/Auto/automake/automake/tests/../lib/am/depend2.am: 
am__fastdepCXX does not appear in AM_CONDITIONAL
 Tom> FAIL: cxx2.test

 Tom> The problem is that we get the missing AM_CONDITIONAL error before we
 Tom> get the missing AC_PROG_CXX error.  The latter is much more
 Tom> informative to the user, so just leaving things as-is isn't the best
 Tom> idea.

We have the same issue with the `AMDEP' conditional.  Its
absence is always mentioned before Automake complains about a
missing CC/CXX/GCJ...  I've thought about introducing a
&require_conditional function much like the &require_variables
function I added recently.  However a better fix would probably
be to arrange Automake so that it complains about a missing CC
before AMDEP.  I don't know how easy it is.

[...]

 Tom> if %?LIBTOOL%
 Tom> ?GENERIC?%EXT%.lo:
 Tom> ?!GENERIC?%LTOBJ%: %SOURCE%
 Tom> -if  %AMDEP%
 Tom> +if %FASTDEP%
 Tom> +## In fast-dep mode, we can always use -o.
 Tom> + if %COMPILE% -MT %LTOBJ% -MD -MP -MF "%DEPBASE%.Tpo" \
           ^^^^^^^^^
This should probably read %LTCOMPILE%.

-- 
Alexandre Duret-Lutz




reply via email to

[Prev in Thread] Current Thread [Next in Thread]