Le 29/05/2012 13:50, Marcel de Rooy a écrit :
I agree with most responses in this thread: If a company makes a patch, a customer of that company signs off, it should not be QAed by that company, but by a "neutral" party. The QAer should even be allowed to ask for a second outside signoff if he feels the patch needs that additional proof.
What's the QA done for ? it's looking at the quality of the code. So, a QAer can request a 2nd signoff, but only if he feel that the code is missing a case (like "I feel this code will work for UNIMARC, could someone check for MARC21" ? or "work for sysprefX = OFF, must be checked for sysprefX = ON as well". I feel that should be an uncommon case. The RM has a more global responsibility to ensure the global consistency of the soft.
As Paul mentioned earlier (he does QA but does not set the status): In order to prevent the appearance of pushing the process, you could even ask if it would be wiser to refrain from such QA comments. (Or just mail them to the author.) I'm not sure I understand ? Do you mean I should not QA if I can't set the status ? Sound a very bad idea : if I see something wrong, like an unconditionnal warn, a missing use Modern::Perl, it must be said publicly, to avoid having another QAer requesting this problem to be fixed
Reminder (or information in case you don't know) = I usually QA & push patches ordered by last modification date, ASC (So the oldest 1st). If I see something wrong while I'm reviewing, I don't understand why I should stay silent ! HTH -- Paul POULAIN http://www.biblibre.com Expert en Logiciels Libres pour l'info-doc Tel : (33) 4 91 81 35 08