On Sun, 12 Jan 2003, Andrew C Aitchison wrote: >> As far as commit access goes, frankly, if those asking for it could >> establish a record of submitting complete and correct patches that didn't >> need review (and Mike, your record on this isn't anything to boast >> about), then you might have a better shot at it. > >We *need* more developer-hours on the commit process.
That would be a nice thing to have. It works marvelously for all other projects that have wider spread CVS write permissions. >Maybe we have to put up with people like Mike making a few more >mistakes while that they learn what they are doing wrong? I can't agree with you more. I, like any serious developer, certainly want to be told about my mistakes, so I can correct them, or have an opportunity to explain why I believe my changes to be correct. There's only one patch that I can think of off the top of my head which I've sent to XFree86.org which was rejected, to which I later resent an updated version back with a good explanation for why I thought it should be applied. I don't recall receiving any feedback positive or negative since. It's not a major issue however as I apply the patch to our sources regardless, as I believe it to be correct, and it harms nothing at all. Most likely I will post that patch here again in the future - now that the development list is publically open. I am then going to get much wider peer review of my patch, and instead of getting one person's opinion of it's correctness, I can get 100 people's opinion. While it may not make the patch accepted into the upstream official source even if 500 people agree, it would at least make it a case 500 separate opinions, instead of the prior case of one person's opinion in a position of power's versus one person's opinion whom is not in a position of power. I'd rather be judged by 500 peer developers for the quality of my patches and work than by one single person. >They will learn a lot faster by doing it wrong than by submitting >patches that may or may not appear months later in a very different >form. Indeed. I will say that a while back, I approached David concerning patches and what his expectations were. I have made every effort to try and meet those expectations, and have modified some of my ways of doing things further beyond what was requested, in an attempt to provide the project with better submissions. In the past, I would not only send in my own patches (which incidentally get applied unmodified almost always), but I used to submit many patches from others, usually - but not always with proper attributions on them. Some patches you find in the wild with no indication of whom the original author is. I just passed the stuff all on upstream hoping that someone there would review them for correctness, apply them, or reject them, etc. Some patches I just picked up of mailing lists as I saw them, and submitted them so they wouldn't get lost. Several of these patches as you can imagine, were not the highest quality, and many never got into the tree. Unfortunately however, many of them were wrongly I believe attributed to ME, but of no fault of my own for not being more organized with what I sent in, and without as much process and procedure in place as what I have now. I also used to send in patches that people sent into Red Hat via bugzilla. We receive many patches of which I personally am not familiar with the particular area of code that is being patched, and so I am not always personally capable of vouching for a given patch's correctness. In such cases, what should one such as myself do? Tell the person to send their patch to XFree86.org instead? People get PISSED BIGTIME! I've told people to please submit xkb patches (MANY of the ones recently sent in to [EMAIL PROTECTED] which subsequently were applied to CVS, and if need be I can provide bugzilla bug ID's of proof). People go nuts! "I am a Red hat user and you should apply my damn patch blah blah." If I send it to XFree86.org myself, then am I being judged as having vouched for it's correctness? I didn't think I was before, but after David's comments toward "my patches" before, I took a hard line about it. For a few months I have refused MANY patches submitted, and requested the person to send their patch directly to [EMAIL PROTECTED], and that if it were accepted upstream, I would apply it to Red Hat Linux also. I also have stated to people when doing this many times (and this is also bugzilla queryable for the conspiracy theorists amongst us) that by submitting the patch upstream, ALL distributions get the benefit, as does the project as a whole, and that it doesn't end up being a fix that is only seen by Red Hat users. >As others have said, feedback on problem submissions is important; >does it add much time to fire off a quick note at the point where you >find a problem with a submission ? You don't need to tell us the >solution. If something is wrong with a patch I submit, or a fix I make, I _WANT_ to know about it, and preferably by MANY people - not just one. >Without access to the submissions* I can't tell, but maybe if we >we would make better progress with 99% accurate commits than >insisting that every one was 100% perfect (90% would *not* be good >enough) ? Sure, that way, if I submit a patch, and it sucks, I can get a kick in the head from 500 people instead of only a few. I want to have numerous people overviewing anything I submit, and I'd like also to have just as many people reviewing other people's submissions too. Code quality is important. >I'd not in a position to accuse anyone, but I'm reminded of the saying: >"The best is the enemy of the good". > >* Read access to the patch@ and fixes@ lists would be helpful, then >we would all have an idea of the backlog. Very much so. Since David announced the other day that all of the previously private members-only lists were going to be made public, I totally give him the benefit of doubt if these were overlooked, that it was an honest goofup to forget to make them also public, since they're not mailing lists in the more traditional sense. Since I am a member however, I get both of these lists automatically and can't be sure if they're public now or not. David, could you comment on this? -- Mike A. Harris _______________________________________________ Devel mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/devel
