Re: RFS: gaim-encryption -- encryption plugin for gaim
Hi Leo,
Well, first let me say that I didn't know about your ITP [1]. I just
saw now that your page is the one linked from the gaim-encryption
website, that I actually had seen before.
I decided to make my own gaim-encryption package, since:
1) I thought it would be something useful to have in the main Debian
repository (remember that I did not know that you had any intention in
officially packaging it)
2) I really wanted a plugin package and not a combined package. I have
several gaim-plugins and don't want to have one gigantic package with
all possible plugins... then there would not be much advantage to the
pluggable architecture, would there?
You're right about the gaim headers: it would be much nicer to have a
gaim-dev package (that's why I also wrote to the gaim maintainers).
But, since it is not there and it won't be there in the near future
we're better off being pragmatic and transition to the ideal solution
when it's possible. Perfection is nice, but I rather prefer to have a
pragmatic but working solution than none.
In response to your three points:
1) only when the gaim plugin APIs change (and they should be backwards
compatible throughout the 1.x.y versions), so that is not such a big
problem
2) right, in an ideal world
3) no, see further
When I made the package, I came to the conclusion that the missing
gaim-dev is not the main problem. The compile options used by the
Debian gaim maintainers were. They namely used --disable-nss, which
breaks gaim-encryption.
I contacted both the gaim maintainers as the gaim-encryption developer
about this. The gaim-encryption developer was very helpful and adapted
the upstream program so that it now initializes nss itself when nss is
disabled and makes sure it doesn't get initialized twice when nss is
enabled.
So, no, it is not a Debian-only patch. It is a solution to the
incorrect assumption the nobody would compile gaim with --disable-nss.
This is now in the default upstream release.
Btw, I would have contacted you if I knew you were serious about adding
your package to the main repository... I certainly did not try to
package it behind your back, I did make an ITP and talked about it with
the gaim maintainers and the upstream developer. None of them told me
about your efforts.
So, in short: I still think gaim-encryption is a very useful package to
have in Debian, and no I don't think my package is a 'hacky' one. Sorry
about the bad communication, though, that was my fault... I should have
looked harder for a possible ITP bug.
Cheers,
Chris.
[1] I (incorrectly) thought searching for gaim-encryption on
http://www.debian.org/Bugs/ would give me the ITP bugs too...
---
On Mon, 2004-10-18 at 14:22, Leo "Costela" Antunes wrote:
> Hi
>
> On Seg, 2004-10-18 at 05:06, Chris Vanden Berghe wrote:
> > I've been in contact with the gaim developers who seem to be reluctant
> > to add a gaim-dev package in the near future (they are considering an
> > experimental package, though). Therefore, I think it is better to go
> > ahead and just ship a copy of the gaim headers until the gaim-dev is
> > available. Btw, the original ITP is more than one year old...
>
> Shipping Gaim's headers with gaim-encryption's source is a bad idea,
> IMHO, for three reasons:
> 1:You'll have to sync it with Gaim for every new version and you'll have
> non-working plugins in the meantime between gaim's upload and your
> upload
> 2:Gaim's headers don't belong in gaim-encryption, they belong in gaim.
> 3:Your source will be different from upstream (at least if I understand
> you correctly)
>
> My ITP is more then one year old because I believe any other solution
> that's been suggested this far to be a kludge.
>
> > I've also been in contact with the upstream author who, at my request,
> > has altered his plugin especially for Debian. The Debian gaim package
> > is namely compiled with --disable-nss. Thanks to this change, this
> > version of the plugin can be used without altering the gaim package
> > (before it could only be shipped as gaim+gaim-encryption, now it's a
> > real plugin).
>
> What change was this, exactly? Did he port the whole NSS+NSPR code from
> Gaim to gaim-encryption?
> Has this change been applied to the normally distributed source or is it
> a debian-hack-only?
>
> > I think people would really like a gaim-encryption plugin package in the
> > main Debian repository...
>
> Yes, I can assume that by the number of emails I get about it, but that
> doesn't justify including a hacky package, IMHO.
>
> I'm sorry if I sound rude, but I've been packaging gaim-encryption for a
> while now and would find it polite to be informed about such plans.
> (Yes, I have all that fatherly crap about packages I work on)
> From what I've understood, I don't particularly like your plan, but
> you're free to try to convince me, prove me wrong, or just ignore me and
> upload it anyway.
>
> Cheers
Reply to: