Thread: Well done, Hackers
Unofficially, well done to everybody that submitted patches in time for Commit Fest. The queue has grown significantly over last few days and it's clear many people have been working very hard right up to the deadline. Depending upon how you count it there are around 5 multi-month mega patches, plus more than 50 other features. Very impressive. I count nearly a dozen new names on the list. Congratulations to you. -- Simon Riggs www.2ndQuadrant.comPostgreSQL Training, Services and Support
Simon Riggs <simon@2ndQuadrant.com> writes:
> Unofficially, well done to everybody that submitted patches in time for
> Commit Fest. The queue has grown significantly over last few days and
> it's clear many people have been working very hard right up to the
> deadline.
"Well done"? It seems to me that we are right where we hoped not to be,
ie with a ton of barely-completed (if not self-admitted WIP) patches
dropped on us immediately before feature freeze. Today the commit fest
idea is looking like a failure.
If we actually manage to ship 8.4 within six months, *that* will be
"well done".
regards, tom lane
On Sat, 2008-11-01 at 14:22 -0400, Tom Lane wrote: > Simon Riggs <simon@2ndQuadrant.com> writes: > > Unofficially, well done to everybody that submitted patches in time for > > Commit Fest. The queue has grown significantly over last few days and > > it's clear many people have been working very hard right up to the > > deadline. > > "Well done"? It seems to me that we are right where we hoped not to be, > ie with a ton of barely-completed (if not self-admitted WIP) patches > dropped on us immediately before feature freeze. Today the commit fest > idea is looking like a failure. > > If we actually manage to ship 8.4 within six months, *that* will be > "well done". Work happens in stages. A lot of people have obviously worked very hard and I wanted to make sure their efforts do not go unnoticed. I don't see contributions and effort as a bad thing for Postgres. And CommitFests have helped in reducing the volume of patches that appear at the last deadline, just imagine what it would be like otherwise. If there was a problem anywhere in sight it would be that the development window closes for 6 months of the year, forcing developers to cram their contributions into a tight window. Patches that take many months to plan and write are simply more likely to appear at the last Commitfest of the year. I'm more than aware that much of the work from here rests on your shoulders, and I understand how you must feel. But let's give credit to the people that have worked hard to get to here and then lets give *more* credit to all the people that contribute from here. We can say thank you to each in their turn. If we can positively motivate people we will be able to spread the load rather than have the load fall on you. Believe me, none of us enjoy watching you struggle (capably) with the huge workload and responsibility. I'll do more than I've done in the past to help, both with my own work and others. Thanks in advance. -- Simon Riggs www.2ndQuadrant.comPostgreSQL Training, Services and Support
Tom, > "Well done"? It seems to me that we are right where we hoped not to be, > ie with a ton of barely-completed (if not self-admitted WIP) patches > dropped on us immediately before feature freeze. Today the commit fest > idea is looking like a failure. > If you don't want patches coming at given deadlines then yes the commit fest idea is a bad idea altogether. But what is the real issue? - The core team is too small to absorb contributions or the development is not distributed enough? - There are not enough guidelines or requirements for a patch to make it to the commit fest? - There is not enough QA manpower/test cases to test the patches efficiently? - Lack of roadmap? Hard to guess what contributions are going to come? What are your views on how the process could be improved? If the commit fest does not work, what should we do instead? I think that complaining after volunteer contributors is the best way to not get any more contributions and have nice empty commit fests in the future. Maybe that's the way to go to solve the issue at hand! ;-) manu -- Emmanuel Cecchet FTO @ Frog Thinker Open Source Development & Consulting -- Web: http://www.frogthinker.org email: manu@frogthinker.org Skype: emmanuel_cecchet
> "Well done"? It seems to me that we are right where we hoped not to be, > ie with a ton of barely-completed (if not self-admitted WIP) patches > dropped on us immediately before feature freeze. Today the commit fest > idea is looking like a failure. > > If we actually manage to ship 8.4 within six months, *that* will be > "well done". It looks to me like there are at least half a dozen patches submitted in the last week that are pretty half-baked and fall into the category of "Let's submit something before the deadline for CommitFest, in the hopes of being allowed to finish it later." This completely shafts the process in two ways. First, anyone who gets assigned to review one of those patches might as well not bother, since the author is probably still working frantically on the patch anyway and will likely find and fix a lot of the issues that any review might turn up. Second, since the authors are frantically working on their own patches, they will have no (or diminished) time to review other people's patches, which is the whole point of CommitFest. It seems to me that "Work In Progress" needs to mean "I need some feedback this CommitFest so I can finish it for the NEXT CommitFest" and not "I'd like to make an end-run around the submission deadline." On the other hand, the number of patches that fall into this category is actually not that large as a percentage of the total. A lot of the larger features have been under development for months and have been extensively discussed on -hackers or were submitted for the previous CommitFest. If you could somehow wave your magic wand and get all of those committed or rejected, I doubt the remaining list would be terribly intimidating. I have five patches in for this commitfest but I bet you (tgl) could deal with all of them in an afternoon without breaking a sweat. ...Robert
On Sat, 2008-11-01 at 16:21 -0400, Robert Haas wrote: > > "Well done"? It seems to me that we are right where we hoped not to be, > > ie with a ton of barely-completed (if not self-admitted WIP) patches > > dropped on us immediately before feature freeze. Today the commit fest > > idea is looking like a failure. > > > > If we actually manage to ship 8.4 within six months, *that* will be > > "well done". > > It looks to me like there are at least half a dozen patches submitted > in the last week that are pretty half-baked Robert, Without review, how can we determine the true state of a patch? I would guess that you haven't done a review of all of them yourself to determine that. Your characterisation of other developers seems negative to me and I really do hope unfair. Nobody I know has submitted anything half-baked, but I haven't reviewed any patches yet. Review is about asking for help from your peers, not an admission of crap software. We're all seeking to learn more and be better, I hope. If this sounds like a flame, its not. I just want to be positive about the efforts of so many people who are all working together as a team. There will be some hard, honest reviews and some patches may not make it, so its a good time to say thanks for trying so hard. For myself, I've submitted a patch touching more than 60 files. Am I nervous I got something wrong? Damn right. Is there some horror lurking in there that I'm not saying, but frantically fixing now? No way. That's not the spirit. -- Simon Riggs www.2ndQuadrant.comPostgreSQL Training, Services and Support
Emmanuel Cecchet wrote: > Tom, > > > "Well done"? It seems to me that we are right where we hoped not to be, > > ie with a ton of barely-completed (if not self-admitted WIP) patches > > dropped on us immediately before feature freeze. Today the commit fest > > idea is looking like a failure. > > > If you don't want patches coming at given deadlines then yes the commit > fest idea is a bad idea altogether. But what is the real issue? > - The core team is too small to absorb contributions or the development > is not distributed enough? The list of people experienced enough to review complex patches has not grown as fast as the number of people submitting complex patches. That will probably even out over time, but it might take years. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + If your life is a hard drive, Christ can be your backup. +
Simon Riggs wrote: > > On Sat, 2008-11-01 at 16:21 -0400, Robert Haas wrote: > > > "Well done"? It seems to me that we are right where we hoped not to be, > > > ie with a ton of barely-completed (if not self-admitted WIP) patches > > > dropped on us immediately before feature freeze. Today the commit fest > > > idea is looking like a failure. > > > > > > If we actually manage to ship 8.4 within six months, *that* will be > > > "well done". > > > > It looks to me like there are at least half a dozen patches submitted > > in the last week that are pretty half-baked > > Robert, > > Without review, how can we determine the true state of a patch? I would > guess that you haven't done a review of all of them yourself to > determine that. I think a complex patch that hasn't been discussed on the hackers list recently, and that has deficiencies reported by the author can be safely described as a work-in-progress that needs feedback, rather than committed for 8.4. -- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB http://enterprisedb.com + If your life is a hard drive, Christ can be your backup. +
On Sat, Nov 01, 2008 at 04:21:30PM -0400, Robert Haas wrote: > It looks to me like there are at least half a dozen patches submitted > in the last week that are pretty half-baked and fall into the category > of "Let's submit something before the deadline for CommitFest, in the > hopes of being allowed to finish it later." Supposing this is true, one way to prevent that in future commitfest-driven releases is to send them back as not ready this time, and tell them they'll be 1st in line in the next go-round. It's worth remebering that this is the first release using the commitfest model, so there will be things to learn from the 1.0 attempt. A -- Andrew Sullivan ajs@commandprompt.com +1 503 667 4564 x104 http://www.commandprompt.com/
> Without review, how can we determine the true state of a patch? I would > guess that you haven't done a review of all of them yourself to > determine that. No, I'm talking about the comments that were made when submitting the patch. A few people said things along the lines of "this isn't really finished" or "this has major bugs". > Your characterisation of other developers seems negative to me and I > really do hope unfair. Nobody I know has submitted anything half-baked, > but I haven't reviewed any patches yet. Review is about asking for help > from your peers, not an admission of crap software. We're all seeking to > learn more and be better, I hope. Sorry. I don't mean to be negative, and certainly not about the developers. I'm actually quite impressed by the number of people submitting patches, and I'm really looking forward to some of the proposed new features. My point is just that about 10% of them sound like they aren't actually done. And as I said in my previous email, I don't even think that's the main issue. > For myself, I've submitted a patch touching more than 60 files. Am I > nervous I got something wrong? Damn right. Is there some horror lurking > in there that I'm not saying, but frantically fixing now? No way. That's > not the spirit. I wasn't talking about your patches. :-) ...Robert
> Supposing this is true, one way to prevent that in future > commitfest-driven releases is to send them back as not ready this > time, and tell them they'll be 1st in line in the next go-round. Agreed! But I'll be as happy as anyone if it doesn't come to that. ...Robert
Tom, Robert, Simon, What, are people just on edge because of the US election? It looks to me like the commitfest system is going really well. Of course, we'll see how long it takes to close out 8.4. But I think we're in much better shape than we were for 8.3. We're even in better shape to reject things. --Josh
Josh Berkus wrote: > Tom, Robert, Simon, > > What, are people just on edge because of the US election? > > It looks to me like the commitfest system is going really well. Of > course, we'll see how long it takes to close out 8.4. But I think we're > in much better shape than we were for 8.3. We're even in better shape > to reject things. I have to agree with Berkus here. Compared to how we used to do things, this is quite marvelous. Joshua D. Drake > > --Josh >
Sent from my iPhone On 2 Nov 2008, at 06:57, "Joshua D. Drake" <jd@commandprompt.com> wrote: > Josh Berkus wrote: >> Tom, Robert, Simon, >> What, are people just on edge because of the US election? >> It looks to me like the commitfest system is going really well. Of >> course, we'll see how long it takes to close out 8.4. But I think >> we're in much better shape than we were for 8.3. We're even in >> better shape to reject things. > > I have to agree with Berkus here. Compared to how we used to do > things, this is Yeah, another +1 from here. As I see it we have: - Given developers timely feedback throughout the cycle, so no one has had to wait 6 months+. - Committed new patches in a timely fashion, avoidind bit rot of those and subsequent patches they affect. - Significantly reduced the amount of work we would have previously seen at feature freeze. I would also note that this last commit fest is really feature freeze, and should be treated as such. That means that it's the cut off point for new features which are essentially ready, not that code must be 100% complete and tested. I doubt there are many big patches that haven't require work after freeze. For future clarity, perhaps we should have a feature freeze point, followed a 'polishing freeze' by which time the submitter must finish pre-review cleanups and last minute fixes. /D
On Sat, Nov 01, 2008 at 08:07:13PM -0400, Andrew Sullivan wrote: > On Sat, Nov 01, 2008 at 04:21:30PM -0400, Robert Haas wrote: > > > It looks to me like there are at least half a dozen patches submitted > > in the last week that are pretty half-baked and fall into the category > > of "Let's submit something before the deadline for CommitFest, in the > > hopes of being allowed to finish it later." > > Supposing this is true, one way to prevent that in future > commitfest-driven releases is to send them back as not ready this > time, and tell them they'll be 1st in line in the next go-round. With some feedback, hopefully. I imagine a number of people submit patches because they hope to get some feedback on them, something which does always happen at other times... I think all this development is good. Have a nice day, -- Martijn van Oosterhout <kleptog@svana.org> http://svana.org/kleptog/ > Please line up in a tree and maintain the heap invariant while > boarding. Thank you for flying nlogn airlines.