On Mon, 2006-05-08 at 07:44 -0500, Bo Peng wrote: > I will not have time to maintain a branch.
... > I know scons is not perfect right now, but I do not expect > anyone who does not like it to use it either. Why cannot Lars and > other autotools fans just *ignore* the scons stuff? Because when someone tries to build with scons, and they have a problem, they are going to send an email to lyx-devel about it. If you don't have time to maintain a branch with the scons stuff in, then you can't possibly have time to handle all complaints about it either. > What is so bad about giving people a > choice? What is terribly wrong to let windows developers compile lyx > quicker and easier? Of course nobody's objecting in principle to making life easier for windows developers. But maintaining two build systems is a significant decision, and unless you're going to maintain it yourself, perhaps you should have asked the people who would end up doing so. By the way, I hate autotools. I hate them because they're too complicated. Having scons+autotools is even more complicated. Autotools needs a replacement, not anything extra. If scons is in subversion, then it's out there in public as the official development branch of LyX. Speaking as someone who regularly handles bug reports on lyx-gtk, which is not ready for primetime, but is out there anyway (packaged, even), I can tell you that if it's not ready, you don't want it out there. I think our multi-frontend situation provides additional evidence for the downside of having many implementations of the same thing. Changes in the backend need multiple implementations in the frontend. With scons+autotools, changes in the build logic need multiple changes in the build tools. So let's have a branch, and let's have a serious look at scons as a potential replacement for autotools. John
