status of forked zlibs in rsync and zsync

[Toshio Kuratomi]

On 09/29/2009 05:00 AM, Rahul Sundaram wrote:
> On 09/29/2009 05:14 PM, Josephine Tannhäuser wrote:
> 
>> Seems that violations of the guidelines are not so important like the
>> violation of the Trademark (The hunting of fedora related sites, like
>> blogs or forums with adhesions contracts)...  Are the project related
>> activities are out of balance?
> 
> They are called guidelines and there are always exceptions. Bundling a
> library is not ideal but removing rsync would be a extreme step. I don't
> think the situation warrants that. Let's not loose perspective here.
> 
So in this case, I think the following things could be said:

* Removing rsync is not an option because of how widely it is used.
* Bundling libraries in zsync is not an option
* If there really exists a way to build against the system zlib and
remain compatible (which librsync claims, someone notting talked to
does, but zsync upstream does not) then that should be done and we'll
patch our packages to do so even if upstream does not accept it.
* If we have to use a forked zlib, we'd patch our packages to pull the
zlib out of the rsync package with its own soname and ship that so zsync
and zsync can interoperate even if upstream does not accept it.  However
we need to coordinate this with upstream or the other distributions to
come up with one standard libname, version, etc.

The alternative is that:
* People assert that bundled libraries are not problematical enough in
the general case to warrant prohibition so the Guideline should be
dropped.  To do this probably means adding other procedures for
maintainers to follow to address the issues that do exist with bundled
libraries.

-Toshio

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 197 bytes
Desc: OpenPGP digital signature
URL: <http://listman.redhat.com/archives/fedora-devel-list/attachments/20090929/ab502906/attachment.sig>