Re: 200 packages with the greatest number of patches

[David Holland <dholland-pkgusers%netbsd.org@localhost>]

On Sun, Jul 31, 2016 at 10:32:11PM +0200, Joerg Sonnenberger wrote:
 > > Current top-10:
 > > wip/chromium-new      448
 > > wip/chromium      333
 > > wip/virtualbox      211
 > > mail/thunderbird38      174
 > > security/tor-browser      171
 > > www/firefox38      166
 > > www/firefox31      165
 > > mail/thunderbird31      160
 > > mail/thunderbird24      157
 > > x11/xview-lib      156
 > 
 > That's not a very useful list. The Mozilla core has a bad record for
 > accepting patches and even if they do, there are lots of older versions
 > that needing them. xview-lib is just dead upstream, so not very useful
 > either. Chromium is similar yet different to Mozilla in that upstream
 > has expressed a clear disinterest in anything not Windows, Linux or OSX.

Preparing patchkits (like I have several of) makes the process easier:
the idea is that the patchkit is directly upstreamable (that is, you
keep pkgsrc-specific stuff out of it) plus it also reduces the pkgsrc
overhead.

The downside is that for packages with a live upstream, you *have* to
maintain the patchset as new upstream versions come out or an even
worse kind of entropy sets in.

In the past I've suggested that TNF set up repositories for this kind
of thing so that any developer can do the work. This always gets a
vaguely positive response so maybe it should happen sometime...

The problem with xview specifically is that I started on this, and
then someone else committed a bunch of conflicting changes without
coordinating, and I've never managed to sort it out as it's not
exactly a super high priority. The result of this is that there's
three or four overlapping but not equivalent patch sets...

-- 
David A. Holland
dholland%netbsd.org@localhost