Hi,
When I started to write my version of the script, I wanted a simple=20
replacement for build/build-many which would check for errors during the=20
builds. You may disagree with the way I do it (I don't think it is a grea=
t=20
script either, but it was written very late too). I specifically chose to=
=20
only allow predefined configurations. I think I made it rather easy to ad=
d=20
new configurations, but the main point was to let a user compile sablevm=20
easily. Not all users will be developers, and most of them propably don't=
=20
know a lot about the different configurations. I suspect that most users,=
=20
given the choice, will opt for the fastest version of sablevm, not even=20
bothering to look at what makes it faster. I think my script is not all a=
ll=20
what sablevm-developers want to have, but again this was not the goal.=20
Grzegorz's script is much better for that.
Anyway, I have no objection to merging both scripts, although I think tha=
t we=20
could have one version (closer in spirit to what I have done) in the=20
"official" sablevm package and one (closer to Grzegorz's) in the CVS. Thi=
s=20
way, developers would have what they want, and sablevm users too.
Cheers,
Bruno
On February 24, 2003 07:26 am, Chris Pickett wrote:
> you guys can work something
> out. =A0I think the various sablevm builds should not be built into the
> script itself, but only the different options passable to configure (it
> would be nice to see one letter abbreviations though, like SDIGTB for
> switch, direct, inlined, GC, traditional, bidirectional). =A0 Rather, t=
he
> user can invoke the script specifying how many different builds and wit=
h
> what options. =A0If they like, this can go in a secondary script. =A0Th=
e
> main purpose is to simplify the autoconf routine, as well as handling
> CVS and .tar.gz distributions. =A0I don't know, I'm rambling.
>
> Cheers,
> Chris
|