From: Alexandre Oliva (oliva@dcc.unicamp.br)
Date: Tue Nov 10 1998 - 18:23:31 EST
I wrote:
>> Someone has already posted a patch to automake/libtoolize Kaffe, but
>> it was never installed :-(
On Nov 10, 1998, Godmar Back <gback@cs.utah.edu> wrote:
> Well, Alexandre, don't complain. You have write privileges and
> are familiar with automake. You can implement it, get our feedback
> and commit it if it's convincing.
I didn't have when the patch was posted. It was quite a long time
ago, and I no longer remember who it was, but I think it was some
member of the japhar team. I was for the installation of the patch,
but Tim (or whoelse had write access at that time) never did it :-(
So I thought I wouldn't go ahead and install it now. I didn't save
the patch, anyway...
> Back then, I raised a few concerns regarding availability, portability, etc.
> I'm still not convinced.
Would it help if I told you I'm current in charge for the development
and maintenance of GNU libtool? :-)
> I tried automake with japhar, and hit a bug right away that Tom
> Troney asked me to track down.
I don't remember having read your message in the automake mailing
list. If you can't/won't track a bug yourself, other people may do
it. Tom doesn't have to take all the load for him, and asking for
help in tracking a bug down is not wrong, IMHO; don't blame on him
just because he did it. If you had posted to the mailing list
(assuming you didn't), someone else might have fixed the bug for you.
Or you could have worked around it. I've been using automake for a
long time now, and I did meet some bugs, but none of them was serious
enough that I couldn't work it out easily. And release 1.3 has been
*very* stable...
> (*) This sentence shows something that becomes apparent sometimes:
> the lack of a decision procedure for this project. I see different
> models:
> a) The Linux model where Linus == Tim. Tim decides what patches go in.
> b) A voting model. I think that's how Apache does things. You need a certain
> number of yes votes, and there's also the possibility of vetoing.
> We could restrict the veto right to Tim or something.
I'm used to the egcs model: people post patches and some of the main
maintainers must approve it; of course others can post comments about
the patches.
But maybe we don't need that... Since we have CVS and the project is
reasonably small, it's easy to revert a change, so we might just go in
a install a change if we feel like implementing it before discussion.
I've insisted in discussing every change I'm willing to make before I
make it because my spare time has been very scarce lately, so I
wouldn't like to waste some work on a patch that won't ever be used.
> (**) related, automake requires perl5. Do all architectures that Kaffe
> covers have perl5?
automake does not have to run on every architecture; it only needs to
run on architectures in which Makefile.am/configure.in files are
modified. If automake is missing and its Makefile finds Makefile.in
to be out-of-date, it just cheats and touches it, assuming it was just
an incorrect timestamp problem.
--
Alexandre Oliva http://www.dcc.unicamp.br/~oliva aoliva@{acm.org}
oliva@{dcc.unicamp.br,gnu.org,egcs.cygnus.com,samba.anu.edu.au}
Universidade Estadual de Campinas, SP, Brasil
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:02 EDT