On Thu, Aug 07, 2003 at 03:42:20AM +0200, Guillaume Rousse wrote:
Ainsi parlait Luca Berra :
Oh, i am speaking about contrib packages, packages in main are another
story.
It doesn't infirms my initial point:
- backward compatibility has a cost
- and a revenue
- having just some packages supporting it by packager choice makes no sense
well, maybe not, packagers could accept patches to make x build on older
releases.

otherwise as introducing conditional macros everywhere in all spec files.

Just releasing an updated rpm package for stable release with missing macros for instance.

I can do only the former of the two. You can do the latter (mklibname in 9.0 could help a lot).

Let's put it this way:
- I have to mantain older systems than my cooker box at home
- If I package something it is because I need that packaged, probably on
 more than one system.
- If I have to mantain a compliant package, then hack it everytime I
 must build on a previous (stable) release, I'll shortly get tired of
 contributing.

We could suggest (or require) that compatibility stuff is kept in
evidence at the top of the spec file, and not scattered around.

L.

--
Luca Berra -- [EMAIL PROTECTED]
       Communication Media & Services S.r.l.
/"\
\ /     ASCII RIBBON CAMPAIGN
 X        AGAINST HTML MAIL
/ \



Reply via email to