#archlinux32 | Logs for 2023-02-26
Back
[02:04:14] -!- drathir_tor has quit [Ping timeout: 255 seconds]
[02:06:04] -!- drathir_tor has joined #archlinux32
[08:51:35] -!- abaumann has joined #archlinux32
[08:51:36] <buildmaster> Hi abaumann!
[08:51:36] <buildmaster> !rq abaumann
[08:51:37] <phrik> buildmaster: <abaumann> deep42thought: you cannot break cmake, it's already broken.. ;-)
[08:52:29] <abaumann> I'm experiencing major stability issues with linux-lts 6.1.x, I made a linux-lts515 package in the AUR. Also, I have machines which have dropped hardware drivers (like the IPC Vortex RAID controller) which will not work with a kernel 6.x.x
[08:54:04] <abaumann> The buildmaster does things in non-atomar ways, the database operations, the signing and the moving of packages in the mirror directories can get interrupted and then leave an inconsistent state.. Normally you would do everything outside the database first, have inverse operations for every operation in case the last operation in the database fails with a rollback.
[08:56:36] <abaumann> the inverse operation of a repo-add is a repo-remove, the inverse of copying a file or making a symlink is removing it again
[09:21:07] <abaumann> now I moved all the broken stuff in mirror and get "The following packages in community are missing from the database or vice versa:"
[09:28:22] <abaumann> ah. ok. repo-add and repo-remove has also to be called propertly..
[09:29:18] <abaumann> ..the database is correct and did the operations first, but probably not in one transaction, one step failed, then the mirror got left behind..
[09:30:24] -!- bill-auger_ has joined #archlinux32
[09:45:23] <abaumann> "we" are sane again ;-)
[09:57:28] <abaumann> i686/python: eurobuild3-i486 again a wrong subarchitecture scheduling
[11:20:56] -!- ligma_1 has joined #archlinux32
[12:36:23] -!- pants has quit [Remote host closed the connection]
[12:38:42] -!- pants has joined #archlinux32
[14:38:08] -!- abaumann has quit [Quit: leaving]
[15:19:36] <girls> the inverse of a repo-add might also be a repo-add of an older version
[15:19:43] <girls> it's never as simple, as it looks on first glance
[15:34:52] <girls> theoretically, the buildmaster does the database stuff atomicly and with an intention-log
[15:35:11] <girls> and if it fails to complete something, it "goes insane" and one can manually look at the intentions
[15:35:28] <girls> however, lately I didn't see the intentions lying around after an error - strange ...
[15:46:14] <KillerWasp> i'm stuck in a older version of LTS. It's frustrating when you can't use the GPU, opengl, or wifi connection.
[15:46:59] <KillerWasp> So I just update arch linux on a different partition, but I'm not using it.
[15:48:14] <KillerWasp> The problems that need to be fixed are from the source code of some packages, such as drivers or kernel drivers.
[16:19:33] -!- abaumann has joined #archlinux32
[16:19:33] <buildmaster> Hi abaumann!
[16:19:33] <buildmaster> !rq abaumann
[16:19:34] <phrik> buildmaster: <abaumann> apache is nice.. if it works.. if not, it's terrible
[16:33:15] <abaumann> KillerWasp: I had the same problem, I cannot really port forward the vortec drivers in the kernel.. so 5.15.x LTS it is for that machine.
[16:49:57] <abaumann> mmh, strange, python is broken on pentium4 and i486 but not on i686?
[17:31:09] <T`aZ> and nobody is maintaining a dkms version of that old driver ?
[17:31:43] <T`aZ> i also despire idiots removing perfectly working software, for the sake of 'it's old lol lets remove this'
[17:32:22] <Menzador> If you had your way we’d still be on 2.6.32
[17:33:28] <T`aZ> i disagfree, this is not how software development works, hundreds of hours goes into a driver, maintaining it and tweaking stuff once in a while is not that hard
[17:34:14] <T`aZ> and even though the kernel internals are evolving, it never requires a complete rewrite
[17:36:44] <KillerWasp> T`aZ: i found much programmers that maintain your own dkms, like each one maintain your own tuned car. :/
[17:37:58] <KillerWasp> seem is more easy take a old dkms or other software and add patchs by your own way.
[18:08:29] <abaumann> actually, the driver had some issues and nobody wanted to adapt it to some API changes. And to be frank, the cards are 20 years old, so.. :-)
[18:08:57] <abaumann> Just checked, 6.2 still runs on i486, so no reason to switch to an LTS kernel 5.5.x :-)
[18:10:44] <KitsuWhooa> Speaking of removed drivers, the S3 Savage driver was nuked from the kernel. I'm going to have to try to DKMS it at some point...
[18:11:10] <KitsuWhooa> having no fb acceleration is really painful on such old hardware
[18:31:06] <abaumann> yeah, true.
[18:36:50] -!- lithiumpt has quit [Ping timeout: 255 seconds]
[18:57:57] -!- abaumann has quit [Quit: leaving]
[20:23:43] <T`aZ> KillerWasp: yeah, but it moves the work from sombody that master this driver/framework knowledge, aka the maintainer, to somebody else that is less versed and has less time, it's it's globally less effective that way
[20:31:04] <KillerWasp> T`aZ: Yeah. It's unfortunate, but that's how it works right now. dkms maintainer doesn't want to fix bugs even if you have good patches to contribute.
[21:13:20] -!- epony has joined #archlinux32