Am Mittwoch, 29. März 2006 13:15 schrieb Mike McCormack:
Willie Sippel wrote:
Probably true. But Wine has several problems where a 'real' fix is so very complicated that we won't see something anytime soon, probably for years to come (like the DIB engine, planned for years, but nobody seems to even work on it). In the recent two years, Wine became unusable for many users - I even know some guys that switched back to Windows because of 'fixes' to Wine that render lots of application unusable, due to regressions expected when those fixes were commited. Those 'fixes' should never make it into CVS if there's not even a realistic timeframe for fixing the regressions.
How do you propose that we figure out if there's going to be regressions or not before the patches are committed?
Well, granted, that won't usually work. However, with the WM rewrite, I think it was expected. And the unmanaged window problem with Steam is well known, as is the workaround. I still don't really know what's so hard about the unmanaged window thing, as there are unmanaged windows on Linux, too (eg, XMMS) - but I'm not really much of coder, so I most probably missed something...
Isn't it just better to start with a patch that is "right", but will still show regressions, then fix those regressions, as opposed to starting with a patch that is wrong, and then hacking on it forever trying to solve the unsolvable problems that causes?
You are right, of course. I'm all for doing stuff right, and I'm not a friend of quick-and-dirty hacks myself. I simply can't understand why most serious regressions introduced in the last two years are seemingly not worked on at all - they simply seem to get ignored. I'm sorry if my mail sounded like a rant, but I'm one of the guys obviously lucky enough to be affected by pretty much every single regression over the last two years in some way - and that gets really frustrating. Of all the applications I used with Wine when I switched to Linux a few years ago, only a single one still works, while every single bug this application shows was already in Wine in 2003.
The WM-rewrite/ broken windowed-OpenGL support is a perfect example, rendering
My counter example is richedit. As a half-baked wrapper around the edit control, it rotted in the CVS for at least 4 years until Krzysztof finally rewrote the whole thing. If the half-baked version never existed in the first place we would have had a working richedit control alot sooner.
It might be a little bit harder to get things done by avoiding hacks at the start, but it's *alot* harder to get rid of the hacks once they're there. People will complain and complain forever that the hacky version worked for them once for one application, even if the new version does alot better for all applications.
Correct. It seems to be a matter of motivation. I don't know... An idea would be to accept an evil hack to make unmanaged windows work, with a deadline. Like, the patch goes in for now, but will be removed in three months. So, if someone needs the affected applications working, either write and commit a real fix in the next three months, or the app stops working then.
One could argue that the hack could as well be ommited then 'till someone really fixes the issue, but in case of Steam, the hack is also needed to run and manage Valve games. If nobody can run those applications, nobody can work on bugs they expose as well.