Going for gobjectification

[Jorge Villaseñor]

On Thu, Apr 18, 2013 at 12:40 AM, <gillou.ray at free.fr> wrote:

> Hello,
>
> I’d like to give a try at doing this gobjectification work, and I have
> a few questions regarding this. First let me point out that it’s not
> gonna be a GSoC project because I’m not a student anymore. (And I have
> huge doubts that any student is ever going to do this.)
>
> My main motivation is to eventually bring my detachable sessions branch,
> which rely on a proper gobjectification, towards a usable state. I
> don’t have a deep knowledge of gobject nor libpurple, but I already
> hacked with both to a fair extend. Actually did more gobject/glib than
> libpurple, so I’m not that confident about the actual amount of work.
> In particular, I saw some of you treating gobjectification and struct
> hiding as a rather tedious work, and I’d like to know what actually
> makes it so difficult. Could anyone elaborate on this?
>
> I’ve been told that it would be smarter to start from scratch based on
> the main branch than trying to merge the existing outdated
> gobjectification branch. Anything under cpw.gillux would do? Or some
> non-cpw branch so that others may join the party?
>

The problem with the old gobjectification branch is that there were some
struct hiding work on main which was done differently than the one done in
gobjectification, so now we have A TON of conflicts while trying to update
the branch from main. If you manage to do that merge cleanly I don't see
why not to use it. I guess there was a good amount of work done over there.

We kinda decided to drop the branch because it have been slowing the
development in some way because people have "would this feature break
gobjectification?" in mind and since nobody was working on it, it was
better to just drop it and continue development.

Any of those branches can work, but I would prefer to have a non-cpw branch
for that work.


> Taking lessons from the previous gobjectification work, it seems we
> should really try to merge things with the main branch often, because
> otherwise it becomes a pain to merge and eventually gets dropped,
> right? Or is it a pain to merge anyway because gobjectification tends
> to blow everything up by nature?
>


> What would be the best merge strategy? Gradually merge gobjectification
> work to main (like one PurpleStuff struct after the other), or regularly
> merge main to gobjectification?
>
>                 — gillux
>

Yeah, it would be easier if you merge often main branch to your working
branch (gobjectifications or a new one) so it do not diverge.

As a strategy, you better talk with Grim about it. I would suggest to do
all pre-work (struct hiding, and stuff) in main and do just gobject related
stuff in your branch.

Also, you will need to get us define which modules will change during 3.0.0
(so you don't have to do double work on them) and which are stable as they
are now.

Thanks for working on this. =)

-- 
Masca

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing on usenet and in e-mail?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pidgin.im/pipermail/devel/attachments/20130418/b7dd2621/attachment-0002.html>