On 4/27/07, Kon Lovett <address@hidden> wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Apr 27, 2007, at 6:22 AM, felix winkelmann wrote:
> Hello!
>
> I finally managed to find the time to get chicken running under
> mingw on Windows. This is a setup I can sort of support, so I wonder,
> whether MSVC/Visual Studio support shouldn't be dropped altogether,
> as we don't have a Windows maintainer (who can dig deeper into
> MSVC-related problems, and who is willing to invest the time and
> effort). This would mean that MSVC is not officially supported (even
> if we try to keep the build files still compatible to it).
>
> Any opinions?
Do not support. (Frankly, given the lack of a Windows maintainer I am
pleased that it works as well as it does.)
So far I've traveled 3000 miles from Seattle in the northwest USA to Cincinnati in the midwest. My destination is Winston-Salem, North Carolina. NC is part of "the South" and about halfway up the eastern seaboard. It'll probably be close to 4000 miles of driving by the time I'm done. I have family in W-S and will be using it as a stepping stone for a job search, possibly a national job search. I am on the move. I have no idea where I'm going to end up, or what I'll be doing for a livelihood. I'm writing this on a Pentium II laptop that I've inherited from my sister. Rah rah GMail. A Pentium II + web setup is slow enough that I've had some time to think over the build issues unemotionally, after reading about the Windows build problems that have recently come in.
First the easy part. Felix, if you feel you can sorta support MinGW, then by all means do so. If between you and John Cowan you keep the MinGW and Cygwin builds working, then CMake will always have a decent foothold in the Windows world. The Visual Studio builds may or may not work, but they will never be terribly broken.
Now, the political part. Since we're getting bug reports for Visual Studio, it's clear that people want to use it. "We don't support MSVC" is the wrong message. The correct message is "we can support MSVC if you're willing to do some testing and bug reporting." In a CMake build, the incremental cost of supporting MSVC is quite low.
Also at last count, Visual Studio .NET 2003 (my system) worked just fine. Visual Studio 2005 Express SP1 is what's broken, because it's Microsoft's free cheapass stripped down not-a-real-compiler. Maybe it can be solved, or maybe the answer is "use a real compiler." I do know it's real work to figure out, and I still don't wanna do it. Seems nobody else wants to do it badly enough either. This is open source, so whatever. Once upon a time, I wanted MinGW to work badly enough that I Made It So [TM]. It cost me a man-year. Because I laid out that groundwork once upon a time, it would probably cost a completely ignorant person 1 week to diagnose the VS 2005 Express SP1 problem, and fix it if it can be fixed. Somebody has to actually want to spend that week though. Myself, I could probably solve the issue in 1..2 days, if I had those days. Right now I certainly don't, and it may be quite awhile before I'm willing to make the time.
I just don't believe in this idea of a "Windows Maintainer." Framing it that way, just sounds like "we want 1 guy who's sucker enough to do all the gruntwork for us." I don't think open source should be organized that way, although it frequently is that way. I could be the Autoconf maintainer, frankly, if I wanted to be. I rolled up my sleeves and got my hands dirty once upon a time. I learned what was going on and refactored a lot of it. I had to do it to unify the build systems. I had a goal, something was in my way, so I solved it. Do I like Autoconf? No, not in the slightest. I think it's a complete waste of time. But so are lotsa things in computers. People just have to make decisions about what they want and what they're willing to do to get them. We shouldn't have a "Windows Maintainer" just to relieve everyone of would-be responsibilities. The whole point of a unified build system is you get the benefit of other people solving other bugs on other platforms. It's a distributed approach to open source.
Cheers,
Brandon Van Every