[Koha-devel] Signing-off a patch for a customer

Paul Poulain paul.poulain at biblibre.com
Mon May 28 17:34:59 CEST 2012


Le 28/05/2012 17:03, Chris Nighswonger a écrit :
> Hi Paul,

> I believe we need to stick with the principles we agreed to. This patch
> clearly missed the "approval" of a dis-interested party in its initial
> commit to master. (Perhaps Katrin mentioned this at some point, but I'm
> not sure.) We need to take up the slack here and get a disinterested QA
> on this followup prior to pushing it to master.

If you look at
http://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=6858#c15,
you'll see that MathildeF (from Nimes Public Library) tested the patch
-during the hackfest-, so the initial patch has been "sign-off-ed" by
her. Except she's a librarian and don't nothing about git. That's why
I've setup sandboxes (and would like to improve them), but don't expect
her to do more than this.
(That's the same thing with the follow-up, except it's less clear on the
bugzilla thread)

> I am of the strong opinion that going forward we need to maintain a more
> strict compliance with this principle of dis-interested sign-off/QA.
> Clearly at times one or the other may be impractical, however, one *or*
> the other is always possible.
I mostly agree with you here.

About the idea that i'm breaking our rule, i've reviewed all patches
I've pushed since 3.8.0.
Here is the list of the patches that are "BibLibre only" if you look at git:
* 5a29c39c9fe838ba5af8f7c52937bb094b80d67a, trivial follow-up to the
initial patch
* 2338d02e9eded9dca6fdb6d425e1dc675c7c1a2d, patch signed-off by Nicole,
does not appear because the patch had to be rebased (also tested by
MathildeF with sandboxes)
* b35d34e2ae4713c1f2f524ac565e2623e9f5243a, trivial follow-up to the
initial patch
* 18f21b32c6bfb4caefa00d8aaf650a67d1531d1c, trivial follow-up to the
initial patch
(I don't want to count bug 6858 in this list, because, for me, it has
been tested by a dis-interested -well, in fact, calling the library
"dis-interested" is maybe not the best term, but nevermind ;-) )

My conclusion is that I don't break the rule. When I do QA, I go from
oldest to newest. If I see a patch that is from BibLibre only (which is
already an uncommon case), I do my QA comments, but don't change the QA
status, saying "I'll let someone else from the QA team change the status".

I don't want to break the rule. I just request that 3 identified ppl
from BibLibre can sign-off in the name of the customer (I like the Proxy
term).


> Perhaps it may not fit the desired
> schedule of the vendor, but violation of this principle is the first
> step down a slippery slope.

It's not only a "vendor schedule". It's also for users: if the author
says it works, the library also say it works, fact that is expressed
through "the proxy" of BibLibre project manager, I don't understand
where's the problem !


-- 
Paul POULAIN
http://www.biblibre.com
Expert en Logiciels Libres pour l'info-doc
Tel : (33) 4 91 81 35 08


More information about the Koha-devel mailing list