autoconf-patches
[Top][All Lists]
Advanced

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

Re: cross compiler a.out filename prepends ac_tool_prefix


From: Allan Clark
Subject: Re: cross compiler a.out filename prepends ac_tool_prefix
Date: Sun, 21 Dec 2003 19:13:52 -0800

Kevin Ryde wrote:
> 
> [All messages to the list, not to me.]
> 
> Allan Clark <address@hidden> writes:
> >
> > I don't have the luxury of a GNU compiler that -- if I was intimate with
> > the code -- I could simply change the output filename.  Of course, I'd
> > need sourcecode.
> 
> You should be able to make a wrapper around whatever tool you've got.
> If you publish such a wrapper on free terms you might even be able to
> get a note about it in the autoconf manual.

It's obvious that other compilers are accommodated, but perhaps it's a
case of being more common; otherwise, only a.out would be checked for.


> > It's close to the Solaris or sysv5 types, but "i386cc" creates
> > "i386a.out".  Even checking for "*a.out" works for me, but I do have to
> > alter every ./configure I see, since autoconf's lang.m4 checks for
> > "a.out" but not "*a.out".  Even matching ${ac_tool_prefix}[ab].out would
> > work, matching quite precisely on a.out and b.out on non-cross-compile
> > situations, and of course still matching on cross-compile situations.
> >
> > I can see the rationale of avoiding too many wildcards, though.
> 
> Personally I'd be more inclined to bring weird and wonderful cross
> compilers to a common model, rather than to adapt autoconf to them.

This seems possible with layers of scripting or with the source to those
compilers.

I can see the resistance to bloating code for a rare set of platforms;
this gets a creeping quality into autoconf that would turn people away
from using it.  Sure, it's 20 bytes in a 300k+ script, but everything
counts in large amounts.  Already, I wish more projects used autoconf,
they'd be so much easier to port.

Lacking source code and resisting layers of scripts, I would need to
find another solution.  

What if there was a standard way to avoid this test in a ./configure
run?

Allan




reply via email to

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