On May 25, 2008, at 4:21 AM, Ralf S. Engelschall wrote:
>
>
> Well, rpm5.org _TARBALLS_ are never a problem as they _CONTAIN_  
> Libtool
> 2.2.4. The problem exists only if one is 1. building directly from  
> CVS,
> 2. without using "devtool standalone" and 3. on a platform with an old
> Libtool 1.5.x provided by the system vendor. In all other  
> situations one
> AFAIK doesn't have a problem.
>

All true, but at least misses what I use CVS for.

Using a tarball as a reference forces patch generation and distribution
in order to determine "fixes" that are seen on user platforms.

When multiple reference tarballs, and multiple patches, are involved,
cvs is a more efficient means to exchange information to diagnose
problems and fix issues in situ. Most rpm problems go something like
yadda yadda yadda ... 1000 packages ... using yum ... legacy  
platform ...
and often have to be diagnosed/verified remotely. Sure I try to  
reduce the problem
to smallest possible reproducer first, often unsuccessfully.

That's largely my rpm development model, attempting to sort
whatever random issues are currently being reported on an IRC channel
or in e-mail in a timely and efficient manner.

Using cvs (or any version control system) for distribution is more  
efficient than using
tarballs and patches (aka SRPMS).

73 de Jeff