Hello Robert,
Gaim's Jabber maintainer has already stated that he's not going use
libgcrypt, but instead wants to implement Jabber SSL using NSS (although
I think I disagree with this decision, libgcrypt is becoming very common
and its portability has been improving recently). It does not make sense
to have Gaim linked to two libraries that provide the same feature set.
  
I think you mean GnuTLS. GnuTLS does provide SSL, and needs libgcrypt for low-level.
libgcrypt is really a low-level library. No comparison with the huge NSS.
I meant that you take a copy of it from somewhere else, not write your
own implementation. Not that it matters now, for the reasons given
(above and particularly below, and in the mail Ethan sent) your patch is
unlikely to be merged anyway.
  
Making a copy is not something I like in general :
* It is costly (writing glue, separating what is necessary from what is not, you need to understand both the implementation and the concepts, etc...)
* It is wrong maintenance-wise : while the library will evolve, you will have to frequently check the changes in the mainstream, and backport them to your implementation,
* It is also wrong because it goes against the concept of libraries : Someone already bothered to make a library, maintain it, package it, document it. I prefer to consider that there are existing components (and perharps choice among different implementations), and that we glue all these components together when we need them. The library maintainer keeps his responsability over his work, because he is competent and experienced, and we don't always want to know the insides of the implementation. We only want a working implementation.

It's not *that* bad the patch is not merged. It now exists, and if someone needs it, he is free to apply it. That is what open source is for, choice, isn't it ?

I don't think Gaim wants to support a broken crypto implementation for
the benefit of interoperating with people who use a broken version of a
broken IM client on a broken operating system. They can fix at least a
few of those by switching to Gaim for Windows, but the free release of
Trillian is horribly broken in so many ways.
  
I think I disagree with you.
I don't really like the Windows plateform more than you seem to do. However, let's face it, today, in the real world, desktop means either Windows or Mac OS. (It may change in the future, I sincerly hope so, but that's what I noticed (btw : I *did* in a previous job have a Linux laptop.) ).
Concerning Trillian, I don't agree either. It may be broken (file transfer for instance), but it is enough for basic Instant messenging, and that is what people want. Besides, it offers multiple protocols (MSN / AOL / ICQ are the most used) and for most people it will be better to use Trillian than using ICQ + AOL + MSN (less memory-bloat, etc...)
I don't have any figures, but from my personal experience, Trillian IS used, broken or not.

Interoperability, I think, is the first step towards migration (from Trillian to WinGaim) for basic users.


However, all this is unimportant. I have fullfilled my needs, and have contributed my findings back to the community in case it is of interest to someone.
I do understand that there are reasons for this patch not to go in, and perharps this is wiser.
My mission is over.

Best regards,

Ludovic LANGE