Le Tue, 17 Jan 2012 18:13:30 -0800 (PST), James Buren <[email protected]> a écrit :
> vmiklos: > > I commited a basic set of instructions for it. It may need correcting to get > it right. I can't test it right now. > > bouleetbil: > > I feel like I'm being attacked over this whole matter. Yes, it's my fault. > But I've done everything I could > conceive of to fix the problems I created. Does this count for anything at > all? My main reluctance with > your absolute solution is that it doesn't make any allowances for minor bumps. I don't search who break or not break. I don't have any problem with you. You fix the problem after so it's ok I've already send the same mail when vmiklos was the kernel m8r. > > AFAIK, when vmiklos was doing his minor kernel bumps, he never had to make > any sort of WIP. The ones > I've seen most likely to break shit in base are: glibc, udev, major kernel > bumps, and systemd. The only thing > I can really agree with is that we need more testing for most changes, but > generally I think it's overkill to > require many developers to test every single change to base. This adds more > overhead to getting stuff done. > I know vmiklos have break the reboot with sustemd bump. I've already break dns resolution with glibc bump.etc... We can find some examples. Only users/devels do nothing can break nothing. > I remember just trying to get major kernel bumps done. Just doing an email to > devel about it generally gets > few replies. Generally I might get a maximum of 3 replies within a reasonable > time frame. The last kernel > bump to 3.1 only got a single one from mcirsta. See here: > http://frugalware.org/pipermail/frugalware-devel/2011-October/thread.html#11044 It's not a problem, no replie for me say it's ok for all. > > So, it boils down to my main resistance which is due to the lack of interest > I get from other developers. I got > fed up with getting few if any responses to my WIPs, to the point that I only > wanted to bother with it for very > major kernel rebuilds or similar massive projects. I hate doing it for > changes I perceive to be relatively minor; > where a simple test as instructed to me by vmiklos would probably suffice. > > Your main objection seems to be about the possibility of stealth breakages. I > know that happened here. I tested > the new udev to the extent that I could, but it still blew up in my face when > others had issues. The problem was > rooted in the fact that the new udev broke firmware loading for modules. I > reverted it to the previous working > version as soon as I heard about it. I'm truely sorry, but what else can I > do? Ultimately, no one can guarantee any > changes they make won't bring it to a screeching halt. > > Now that I've had a chance to expression my full feelings and position on > this, what can we do to reach a > compromise? Surely we can find something that we can both accept? > If we add some exceptions we will never use basetesting. For me the only exception can be the security fix. I use only frugalware I don't have any rescue system or multiboot (except for my arm box, don't find the time for hack it) because I've trusted to Frugalware and current was "stable" for some production box. If I should have a system rescue for me it's not interresting. After we can have somme accidents. It's the second time I send a mail for broken base and we would have been able to avoid with a wip. And it's not an x application don't start just my box don't boot. Just one question for a kernel minor bump you should rebuild some modules, initrd ... howto you test this bump ? In current after bump/rebuild. And kernel is the "kernel" you can't test all configuration/hardware. Same thing for systemd you can't test all services.... For me it's the same thing when some devels push some FrugalBuild without build the package and test it before. And I'm not agreed with that. base as the name say it's the system base. We should test it a minimum before bump it into current. I've talking with some devels/users about this rules no base bump directly and all say ok. Now if you and vmiklos say ok for base minor bump. It's not with this way we will have more Frugalware users. base is a lego and only one brick can break it. The only thing I'm just sure I will don't send 3 mails for the same reason. If I should create my boulebase I will create it or switch my production box to stable. I will don't die if I don't have the latest kernel some days after the release. And my geek time is finish I should have a working system. For finish I'm happy you maintain kernel you have added some good features as initrd. And you have some good ideas. (I don't only flame you :D) bouleetbil bye > _______________________________________________ > Frugalware-devel mailing list > [email protected] > http://frugalware.org/mailman/listinfo/frugalware-devel _______________________________________________ Frugalware-devel mailing list [email protected] http://frugalware.org/mailman/listinfo/frugalware-devel
