Op donderdag 08 december 2011 00:58:04 schreef Thomas Backlund: [...] > Sorry for being unclear, I was just pointing out that if we cant agree > on the ml, council has the power to decide.
ah, ok [...] > You suggested to push mariadb in the middle of a big kde upgrade, wich > pretty much means half part would have been built against mysql and the > rest against mariadb... (yes I know it should work tecnically, but...) > And then you suggested that if it does not seem to work we pull it back > out 2 days later -> meaning no real time to test, and messing it up > even more... thats not how you do proper testing/qa... i understand your pov; tbh, after reading mikala's mail, i had misunderstood that 4.7.4 was already completely pushed, and so that it was not in the middle of the big KDE upgrade, but after it. (i've only realized my misinterpretation after speaking with mikala about it) [...] > How does this plan sound ? It sounds good, something like this i had in mind, i hadn't shown the real plan yet, as i just wanted a bit more testing. I only mentioned the worst case scenario there ... (thanks for taking your time on this proposal, it's nicely detailed (i guess i still should take some communication lessons)) but yeah, your plan sounds good, i'll try to make sure everything is in order by then. as a reminder, i'm mentioning that 5.5 branch is _not_ released yet, but has all the fixes already from 5.2 to 5.3 to 5.5 (mysql versioning is a bit odd, in the way that the 5.3 and 5.5 are mostly the same; though after 5.5 versioning should be clearer), however upstream said that the planning should allow to be released as final before februari. at this bzr point, all the code is ready, all the fixes are in; now they are allotting time to fix all the new tests (mostly adapting tests) and build errors in non-standard builds; and also some user-specified patches, such as the ones i submitted from the ones we're still carrying over from mysql. Yesterday, there is discussion going on, on how to do versioning for libclient, since by default gcc exports all; which we have workarounds for. I've presented our workaround, so that they could take also our distribution into account for this matter, so by any chance we'll be able to really clean up our build/patches (hopefully). AL13N
