Hi Jonathan, On Tue, Oct 24, 2017 at 12:25:02AM +0100, Jonathan Matthews wrote: > On 20 October 2017 at 17:17, Willy Tarreau <[email protected]> wrote: > > I'd like to collect all the pending stuff by the end of next week and issue > > a release candidate. Don't expect too much stability yet though, but your > > tests and reports will obviously be welcome. > > Are you still finger-in-the-air aiming for a "November 2017" 1.8? I > don't recall where I saw that quote, but I'm pretty sure it was an > intention mentioned ... /somewhere/!
Yes, I'd like this. Yesterday I finally sorted out one big problem I was still facing on the H2 stuff that managed to prevent me from making progress for almost one week, but now that it's over, most of the work on this part consist in progressively re-injecting the code from my previous branch, so this part is mostly annoying but not really critical. From what I've heard on the front of threads, it should be OK already (they still want to rerun a few tests) and for the cache, the code works but some cleanups are being made on the shctx stuff which will have repercussions on the code so William prefers to publish it after the final update. Overall what I think is that if the merge is simple, we'll merge everything at once, resolve the (hopefully) few conflicts and colateral impacts and issue -dev4 or -rc1 with this depending on our level of confidence. If the merge is really hard (ie another round having to be done on the rest due to locking), we may put everything in separate branches for a while. But I'm really aiming at the former. When I initially proposed the dev plan for 1.8 I mentionned that we'd merge no later than end of september and debug during october and november, but in the end the development has been more isolated and terribly difficult (two h2 implems from scratch and two or three for the threads), explaining why we're merging a bit later but likely with more confidence. > No pressure - just wondering :-) You're welcome. I think in the future we should encourage to push the development trees to a public repo, even if they are horrible, so that everyone can follow the progress (and I haven't been good at this myself this time). We just need to figure how to do this so that nobody is incited to used them claiming an accidental use, and we must as well take care not to discourage shy people from making progress by fear of showing horrible stuff to the world :-) Cheers, Willy
