On Mon, 2008-01-07 at 20:56 +0100, Guido Serassio wrote: > At 19:07 07/01/2008, Alex Rousskov wrote: > >On Wed, 2007-12-26 at 23:18 +0100, Guido Serassio wrote: > > > > > As some Squid developer already know, there is a Squid 2.6 "STABLE > > > Windows" branch in the squid-cache.org CVS repository, SQUID_NT_2_6, > > > based on SQUID_2_6. > > > This allow to easily maintain the native Windows port based on MS > > > Visual Studio from which the distribution binaries are compiled > > > (MinGW and Cygwin support are already in the baseline code). > > > >Would you prefer to add Visual Studio support to the baseline code > >instead of creating a Squid3 branch for Visual Studio releases? Is it a > >matter of adding and maintaining project/solution files? > > The Windows "STABLE" branches are used to generate the native > binaries for Windows of Squid using Microsoft Visual Studio. A > dedicated STABLE branch is needed to warrant source code stability > and isolation/protection. > The support of Visual Studio environment is a bit problematic for some > reasons: > - It lacks the support for configure, so there is a statically (and > manually) generated autoconf.h > - The build process is not based on the standard Makefiles > - Many build steps are done using dedicated .cmd files
The above seem to be unrelated to Squid code itself. It sounds like you need to maintain a set of Visual Studio-specific files such as a static autoconf.h, project files, and .cmd scripts. These files can be maintained independently from Squid source tree (e.g., like Squid web site is maintained now). Am I missing some big reasons why you want to branch Squid code to maintain Visual Studio builds? Please note that I am not arguing against these new branches, just trying to understand why they are needed... Thank you, Alex.
