On 12/13/05, David Ferlier <[EMAIL PROTECTED]> wrote:
> > > For libraries that we have modified, there are several ways to take them
> > > out of the tree :
> > >  - generate a patch that makes a vanilla libs into a customized one, and
> > > apply this patch to a given release of the lib when we build the software.
> > >  - submit the patch upstream to request its inclusion in the lib.
> > >
> > > The best solution, IMHO, is the second one, but it won't always be
> > > possible. The first one requires that we track each lib's so that we can
> > > generate a patch that applies blissfully to any new release.
> >
> > i think too that the second way is the better,
>
> I personally highly prefer the first solution. If you knew what
> modifications we make to the third-party libraries we integrate, you
> wouldn't even ask ;)

As I already said, if you are _that_ far from the pristine sources,
you are actually working on a fork. This is your choice, which has
some implications I am sure you evaluated before going down this path;
I am really not in a position to judge whether merging back to
upstream is feasible or desiderable.

>
> All our modifications wouldn't interest the authors of the project
> because they are sometimes hacks, sometimes diffs to work around the
> project limitations (for example with curl, we added just one function
> to be able to use curl with an already opened socket).

Upstream developers could not be interested but, sticking to the curl
example, did you actually try to write a mail to the curl-library
mailing list?
If using curl with an already opened socket is not supported it could
be for a reason; for example, possible reasons include:
1. no one ever asked for the specific feature
2. someone asked, but no developer was interested in coding it (or had
enough time)
3. it is, for some reason I do not know, a bad thing to do
Whatever is true, they are in the best position to state why this
feature is not there. May be reasons 1 or 2, I think they would be
more than happy to at least evaluate a patch.
May be the third (or somethig else), it would not be a waste of time
because you will know more about why the limitation is there, hence in
a better position to decide how to proceed thereon.

Please note, I am not here to tell you how to manage your project. I
am only trying to help identify issues that would prevent the
inclusion of this program in distros (and you have already a direct
interest from 2 of the major ones...)

Gianluca
_______________________________________________
Wengophone-devel mailing list
[email protected]
http://dev.openwengo.com/mailman/listinfo/wengophone-devel

Reply via email to