autoconf
[Top][All Lists]
Advanced

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

Re: AC_REQUIRE problems


From: Stepan Kasal
Subject: Re: AC_REQUIRE problems
Date: Wed, 16 Feb 2005 15:34:39 +0100
User-agent: Mutt/1.4.1i

Hi,
  I haven't read Dan's mail yet, but I want to reply to this:

On Tue, Feb 15, 2005 at 05:09:17PM -0800, Paul Eggert wrote:
> [...] I'm now seeing many cases where it takes longer to run
> "configure" than to run the following "make"!  It's getting ridiculous.

no surprise.  It takes much time to compile zillion small programs to
verify that the system works as expected, eg. has all the system calls
we need.

Then, we compile a few source files which use these system calls.
No surprise it's much faster.

A distribution builders can speed things up by pre-setting the cache
variables.  (But we cannot use a central cache by default, it would bring
many problems.)

Well, the basic pronciple of autoconf is that we don't believe in
conformances to various standards; instead, we check each individual
feature by a separate test.  The fact that configure takes some time is
inherent consequence of this design principle.

Fortunatly, current computers are fast.  Unfortunately, someone has told
this to the gcc folk.

An idea:
I don't believe it's worth the work to have configure to run the individual
tests in parallel.  If someone is able to invent a way to get several
independent results from one run of compiler/preprocessor, that would
help substantially.  (Even if the trick works for certain compilers only,
it would be worth it to put the two alternative methods to the generated
`configure'.)

Regards,
        Stepan




reply via email to

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