Mailing-List: contact new-httpd-help@apache.org; run by ezmlm
Precedence: bulk
Reply-To: new-httpd@apache.org
Message-ID: <008f01bf9e44$45860900$6fa0e320@apache>
From: "Bill Stoddard" <stoddard@raleigh.ibm.com>
To: <new-httpd@apache.org>
References: <4.3.1.2.20000403212316.00ad9d20@pop.ma.ultranet.com>
Subject: Re: Massive changes to Win32 generated MAK files
Date: Tue, 4 Apr 2000 10:44:23 -0400
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit


> The following three commits by Bill & Bill from earlier this afternoon
> involve addition or removal of approximately 3400 lines in generated make
> files, compared with the addition or removal of only 80 lines of DSP.
This
> is exactly the sort of change that would be eliminated by building Apache
> from the DSW and DSPs instead of the MAKs.  Other than in APR, the only
> thing that changed between versions is the headers on which the ApacheCore
> source files are dependant.  This change required no modification to the
> DSP from which the MAK was generated in all but Aprlib.dsp, where the
> length of the MAK changes was >33 times the length of the DSP changes.
>
> Does anyone still object to William's suggestion about the removal of the
> MAK files from CVS, and the modification of the Makefile.win to use the
> command-line-only mode of msdev.exe and the DSP files to build Apache on
> Windows instead of using nmake.exe, and the generated MAK files?
>
yes, I had this thought too when I saw how ridiculous the makefile diffs
were.

I seem to recall some bugs in the VC++ V5.0 msdev that could cause us
problems. Can't recall the details though. Assuming there are no showstopper
bugs, I am +1 on eliminating the Win32 project makefiles and updating
Makefile.win with direct calls to msdev.

Bill