RE: [cp-patches] [Patch] distcheck fixes for generate-locale-list.shand LocaleData.java

[Mark Wielaard]

Hi,

On Fri, 2004-12-24 at 08:08 +0100, Jeroen Frijters wrote:
> Michael Koch wrote:
> > As discussed on IRC this is no solution for Jeroen.
> 
> If there is a good reason not to add it to cvs, that's ok too then I'll
> just add it to IKVM cvs. I just assumed that there wasn't any good
> reason not to, but maybe I'm missing something?

In general we don't like having generated files in CVS. You have to be
very careful to regenerate them correctly or your source files and the
generated source files get out of sync. (Like with the generated jni
header files, which are only checked in in CVS since they can only
correctly be generated with gcjh also from CVS, which is not easily
available. After gcc 4.0 is out I would like to remove them since they
always seem to get out of date somehow.)

If it can be added in a way that makes it kind of easy to make sure it
can also be normally automatically generated by the standard build
(configure --disable-regenerate-files for who doesn't have a standard
posix/gnu shell) then we can probably add it to CVS if that really helps
people. I just didn't have the time to come up with a clean build
solution for this.

This is only possible for files that only depend on other source files
already in CVS and that always produce the same result. So you don't
have spurious CVS diffs, which is the current problem with using
different versions of kaffeh, gcjh, etc for JNI header files. And files
like Configure.java which depend on your build/configuration environment
cannot easily be "pre-generated" since they would differ for each
user/system/install.

BTW. GNU Classpath should compile out of the box on a windows system
with cygwin. If that doesn't work then that is a bug. I would recommend
people that don't have a posix/gnu environment and that want to hack on
GNU Classpath CVS to install cygwin or a similar package so the build
scripts just work.

Cheers,

Mark

signature.asc
Description: This is a digitally signed message part