Re: libc6_2.0.7 release notes...
Manoj Srivastava <srivasta@datasync.com> writes:
> Lalo> as already discussed, the use of epochs alone (of course they
> Lalo> could be combined with something else) would "never stop",
> Lalo> increasing on each pre/upstream pair.
>
> I am tempted to say, so what?
While technically I agree with you (and I'm perfectly happy if we just
decide to use epochs for this), I can also see the aesthetic
objection. Epochs work, but they're not the most
reefined/discriminating tool for the job. It's like using industrial
epoxy when a thumbtack would have been ideal.
> I think this is a valid (and for me, a show stopper)
> objection. We should not pretend to have the real release sources
> when we don't really.
Yes, as far as I'm concerned any proposal that puts stuff in the
Debian revision is dead.
> Lalo> .99:
>
> Lalo> this works, of course. Except if there already _is_ 2.0.7.99 -
> Lalo> what if, after 89723234 releases of gimp 0.99, they decided to
> Lalo> release a "gimp 1.0pre1"?
>
> Huh? Why cant you tack another .99 to it? like gimp
> 0.99.0.99.99.99.99.99.99.99.99.1 ? This method continues to work
> irrespective of what the upstream version is.
The only think I don't like about this is that it's ugly, and it
doesn't make it easy for someone just reading the version to know what
it corresponds to upstream. To alleviate this, you could go with
something like 4.12.99.4.13pre1-1, but if you're going to do that, IMO
it's clearer to say 4.12-4.13pre1-1. I know I objected to it
originally, but now that I know that dpkg can handle the two dashes in
the upstream version (and in fact policy allows it), I like this
better. That's what I've done for now in libgc4.
> Require changes to dpkg, and hence not likely to happen in a
> hurry. Where the pre epochs go is irrelevant; The critical idea
> presented begore was a special numeric tag added to a version that
> makes the version sort *LOWER* in dpkg's algorithm. Unlike epochs,
> which make the version sort *HIGHER*.
I think we should go ahead (unless anyone has a major objection) and
implement with the "~ special sort character idea".
libgc4_4.13~pre1-1 is much better than either of the two other options
listed above.
--
Rob Browning <rlb@cs.utexas.edu> PGP=E80E0D04F521A094 532B97F5D64E3930
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: