Georg Baum a écrit :
Abdelrazak Younes wrote:
Bo Peng a écrit :
What do you think?
Great Job!
I am convinced and I will certainly use with minimal capability
(building LyX is enough).
I fear you both underestimate the amount of work that is needed to replace
the auto stuff. After all, there is a reson why it is so complicated.
I don't think we underestimate it. But I really think that the auto
stuff checks for _way_ two many things. Now that we are using portable
library like boost most of the test are really not needed at all. So
porting only what make sense on a case by case basis should not be too
much work IMHO.
IMHO, a really worth project before we get rid of scons is to get rid of
most dependencies of config.h in current source code.
Aopart from what a regular developer uses, there are things like "make
dist" and "make distcheck" that need to work.
Let me restate that a build system in parallel to the existing one does not
make sense. It soon gets outdated, and if not it costs double maintenance
work.
IMHO you overestimate the number of files that get added to LyX source
code, I reckon it is not more that once a week.
If I where you I would rather try to get the existing build system work for
your platforms. I agree that the auto stuff is not nice, but this option
will be far less work than replacing it with something else.
We don't have enough man power for such a big project. Of course I can't
(and don't want) to force you to stop working on this, it is just my
opinion that you could be more useful for the whole project if you continue
with bug fixing and implementing new features.
Bo is interested in learning Scons so I'd say he knows what he is doing ;-)
Abdel.
