[quagga-dev 5541] Re: ospf network command w.r.t unnumberd interfaces

paul at clubi.ie paul at clubi.ie
Wed Jul 2 03:02:14 BST 2008


On Tue, 1 Jul 2008, Joakim Tjernlund wrote:

> I would expect some of these patches to be applied, I am getting a 
> bit frustrated that nothing I sent gets applied, not even the 
> simple stuff such as the SNMP cross compile patch.

Well, a lot of people who submit patches unfortunately don't supply 
diffs to relevant ChangeLog files (which will become the commit 
message). Even for a trivial patch, writing the changelog takes a 
minute or two, for bigger patches, it takes much longer. This 
obviously increases the drag co-efficient for a maintainer to apply a 
patch.

Next: Unforeseen errors and problems are, sadly, standard fare with 
the software 'alchemy' we have today (some optimists call it 
'engineering' ;) ), but it's nice to avoid introducing problems that 
a quick test would reveal. So where a submittor indicates a patch as 
being off-the-cuff and not terribly well tested, this obviously 
shifts the burden of testing on to maintainer. This clearly is 
another drag factor, particularly as the maintainer may not be 
terribly familiar with the setup needed to test the change.

The testing issue weighs particularly heavily at the moment, as we 
desperately need to avoid introducing regressions into 0.99, because 
we really need to turn over 0.99 into the stable series..

We can surely some improve things, and hopefully will. CVS is very 
creaky, and the contributor/integration work-flow with CVS involves 
too many round-trips and idle-waits, compared to more modern toosl.

However, it's not all bad tools and lazy maintainers - there are 
things that contributors could do to ease patch application. These 
things might even be documented in the 'HACKING' file, under 'PATCH 
SUBMISSION'. ;)

regards,
-- 
Paul Jakma	paul at clubi.ie	paul at jakma.org	Key ID: 64A2FF6A
Fortune:
Watch your mouth, kid, or you'll find yourself floating home.
 		-- Han Solo



More information about the Quagga-dev mailing list