apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wesley W. Garland" <wes.garl...@gmail.com>
Subject Re: Modifying Win32 default optimizations?
Date Sat, 14 May 2005 15:36:21 GMT
Has anyone ever considered building APR/Win32 via GNU autoconf and GNU
make? I don't think dictating *those* tools would be ridiculous,
although they might alienate some click-centric developers.

For what it's worth, I've built a number of "UNIX" projects as Native
Win32 and WinCE (yes, really) binaries with GNU make, using VC5 and
eVC3. Including Apache 1.[23] for WinCE. I run the build from Cygwin,
but produce Windows native binaries.

I haven't written any autoconf scripts for Win32, but I'm pretty sure
I could dig up my old Win32 Makefiles if there was any interest. These
work by including a file which re-writes the default rules (implicit
rules) to work with CL.EXE and whatnot.

Actually, I'm pretty sure I could run ./configure under Cygwin,
generate a set of APR Makefiles and with minimal Makefile hacking get
it build VC5, and maybe VC6 (have to check if I have a spare license
for the latter) binaries.  Then, if there is an autoconf wizard on the
list, a configure-with-VC configuration could be spat out relatively
easily (but I'm not that wizard).  Afterwards, rules for the Borland
command-line compiler could also be added.

Just food for thought. If there's somebody out there willing to finish
(the autoconf magic), I'm willing to start by generating nice, clean
GNU make files.

There is one more advantage to this build technique, BTW - it means
you can also use flex, bison, m4, etc, under Windows... basically, an
identical build environment for all platforms, using free tools, but
producing NATIVE binaries.

IIRC (and it's been a while!), the only differences between the UNIX
and Windows versions of my Makefiles for those projects (other than
the include file) was that Windows files needed a rule to produce .res
files. And I just wrapped that in a conditional..

Incidentally, does APR produce MIDL/ODL for it's libs? If not, is
there any interest?  I haven't done a lot of Windows hacking, but I do
have some expertise there. (99% of my Windows hacking is: Here, Mr.
Windows guy, take the library I've ported and this typelib I've
generated, and wrap some GUI goodness around it).


On 5/12/05, William A. Rowe, Jr. <wrowe@rowe-clan.net> wrote:
> [I'm answering specifically to apr@ - so that there isn't so much
> noise on so many lists :)]
> At 12:19 PM 5/12/2005, you wrote:
> >On the specify compiler version used in builds, I think APR should
> >compile and pass tests on VC6, 7, 7.1 and 8.  I've submitted patches
> >that also get it compiling and somewhat functioning on Borland C++
> >5.6 and MinGW.  I don't think that APR should dictate a specific
> >compiler that should be used by the people who bundle APR in their
> >application and APR doesn't appear to provide their own binary
> >distribution.
> +1, we are on the same page.  I would be irritated by a product
> which installed a non-ASF, incompatible binary into /win/system32
> for libapr.dll, but effectively you are right, however they like
> to build/distribute, apr libs in their local tree should simply
> work no matter which compiler.  Even VC5 is fine to this purpose.

Wesley W. Garland
Director, Product Development
PageMail, Inc.
+1 613 542 2787 x 102

View raw message