#archlinux32 | Logs for 2021-06-18

Back
[00:00:36] -!- Alina-malina has quit [Ping timeout: 244 seconds]
[04:56:46] -!- Alina-malina has joined #archlinux32
[05:01:40] -!- Alina-malina has quit [Ping timeout: 268 seconds]
[07:06:58] -!- bill-auger has joined #archlinux32
[08:04:15] -!- sunshavi has quit [Read error: Connection reset by peer]
[08:08:50] -!- sunshavi has joined #archlinux32
[08:55:17] -!- drathir_tor has quit [Remote host closed the connection]
[08:58:12] -!- abaumann has joined #archlinux32
[08:58:12] <buildmaster> Hi abaumann!
[08:58:12] <buildmaster> !rq abaumann
[08:58:13] <phrik> buildmaster: <abaumann> manually updating certificates is so.. well 1980.. :-)
[09:08:09] -!- drathir_tor has joined #archlinux32
[09:08:28] <abaumann> deep42thought: yes, the qca problem disappeared, thanks :-)
[09:08:40] -!- jonathon has quit [Remote host closed the connection]
[09:08:52] -!- jonathon has joined #archlinux32
[09:18:58] -!- drathir_tor has quit [Ping timeout: 252 seconds]
[09:30:03] -!- drathir_tor has joined #archlinux32
[10:30:36] -!- jonathon has quit [Remote host closed the connection]
[10:31:22] -!- jonathon has joined #archlinux32
[10:48:12] -!- abaumann has quit [Quit: leaving]
[11:03:52] -!- mistersmee has joined #archlinux32
[11:10:49] -!- mistersmee has quit [Quit: #fsf-india]
[11:13:23] -!- mistersmee has joined #archlinux32
[11:26:09] <mistersmee> Hey, can we push ncmpcpp from staging to testing and then to stable?
[11:26:19] <mistersmee> Because it is currently broken
[11:26:45] <mistersmee> It's looking for boost-libs 75, while boost-libs got updated to 76
[11:27:03] <mistersmee> I manually pacman -U'd the staging package and it works fine
[11:33:07] -!- mistersmee has quit [Quit: WeeChat 3.2]
[11:54:50] -!- mistersmee has joined #archlinux32
[11:58:03] -!- mistersmee has quit [Client Quit]
[11:58:14] -!- mistersmee has joined #archlinux32
[12:00:02] -!- mistersmee has parted #archlinux32
[12:01:36] -!- drathir_tor has quit [Remote host closed the connection]
[12:02:53] -!- drathir_tor has joined #archlinux32
[12:31:49] -!- n0tiz- has quit [Remote host closed the connection]
[12:32:25] -!- n0tiz has joined #archlinux32
[13:42:20] -!- titus_livius has joined #archlinux32
[14:13:28] -!- abaumann has joined #archlinux32
[14:13:28] <buildmaster> Hi abaumann!
[14:13:28] <buildmaster> !rq abaumann
[14:13:29] <phrik> buildmaster: <abaumann> sometimes I get the feeling that the only thing developing fast on Linux is the number of new bugs.
[14:13:47] <abaumann> mistersmee: sure thing, just pushed ncmpcpp
[14:14:23] <abaumann> there might be more packages needing a push, boost was published as 76, boost-libs was on 75, so I had to force push boost-libs 76, eventually breaking packages along the way..
[14:31:31] -!- pyfisch has parted #archlinux32
[14:40:31] <T`aZ> how hard would it be to allow multiple versions of the same package ?
[14:45:12] <abaumann> there are multiple version all the time: qt5 and qt6
[14:45:29] <abaumann> which package did you have in mind?
[14:45:54] <T`aZ> boost, for example
[14:46:30] <abaumann> ah, yeah, actually there something like boost75-libs or so would be handy, similar as for icu
[14:47:07] <abaumann> but usually we should try hard to use the newest versions where possible
[14:47:13] <T`aZ> i mean, the whole (we need to rebuild everything because there is a so bump) , and breaks everything thaat did not get rebuilt, including aur & third party software is not good
[14:47:30] <T`aZ> i know it's a design issue of arch itself, i was just wondering how hard woudl it be to tweak that design
[14:48:32] <abaumann> not hard, it's more like a policy: "before updating boost, make a boost75 package'.
[14:48:58] <abaumann> but I don't think there is really a need for this upstream: the repos are correctly rebuilt..
[14:49:14] <abaumann> ..and the AUR is breaking also upstream and packages have to be rebuilt.
[14:49:19] <T`aZ> indeed, except for aur & third party
[14:49:20] <abaumann> It's by design, basically. :-)
[14:49:25] <T`aZ> yes
[14:49:59] <abaumann> It's ok IMHO that things break in a bleeding edge/rolling distro.
[14:50:18] <T`aZ> it's not imo
[14:50:44] <T`aZ> at least not that kind of breakage, if a given binary was working at one point it time, it should stay working forever
[14:51:18] <abaumann> well, this would mean always backwards compatible ABIs or static linking.
[14:51:21] <T`aZ> the kernel is assuming this responsability
[14:51:27] <abaumann> Or multiple shared libraries..
[14:51:34] <T`aZ> libc is assuming this responsability, however no distro is
[14:52:07] <T`aZ> yes multiple shareds libraries
[14:52:32] <abaumann> the proper way would indeed be to have a boost57 for lib.so.57 and have one version per major version.
[14:52:38] <T`aZ> i need to take a look at nixos
[14:52:44] <T`aZ> yes
[14:52:52] -!- mistersmee has joined #archlinux32
[14:52:54] <abaumann> or maybe also guix.
[14:53:27] <abaumann> transactional updates are a neat thing :-)
[14:53:38] <T`aZ> yes, but this is another kind of problem :p
[14:54:00] <T`aZ> once upon a time, i has a LFS workstation
[14:54:12] <T`aZ> *tons* of the same libraries installed
[14:54:26] <T`aZ> zero issue to run any binary from 10y ago
[14:54:40] <T`aZ> just that there was no package manager, so it was a mess
[14:55:34] <abaumann> running always contradicts the idea of security updates..
[14:56:11] <abaumann> so you have all local libraries and they make your binaries work, but now you have a vulerable version of a library eventually.
[14:56:15] <T`aZ> let the user take the responsability of running insecure software
[14:56:52] <abaumann> the current release mess in Linux is exacly a consequence of moving the responsability to the user/developer IMHO
[14:59:04] <mistersmee> abaumann: yep, I think most of the packages in stable that depend on boost-libs might be broken. Some like blender have correct rebuilds that are in testing.
[14:59:55] <abaumann> oh dear..
[15:04:39] * abaumann reschedules the list on https://archlinux.org
[15:08:56] <abaumann> T`aZ: there is another thought, libraries should be longer more stable, then there would not be so much breakage. boost and icu are too big so every new release triggers an ABI change, another option is to have an ever growing library just with func1(), func2_() for ever function.
[15:10:31] <T`aZ> im fine with having 45069 versions of the "same" libraries
[15:10:49] <T`aZ> this is how the other os is doing for directx
[15:11:40] <abaumann> "the other os which should not be named" ;-)
[15:11:53] <T`aZ> you have one release per month, you link your binary to that specific version that you are using
[15:12:06] <T`aZ> everybody has also that same specific version, you know it wont break
[15:13:09] <abaumann> When the software you install needs 10 different Microsoft C/C++ runtime Libraries (the author has been there).. :->
[15:15:33] <abaumann> Unix was supposed to be a source code distribution originally, so there was nothing wrong to 'rebuild world' all the time..
[15:16:55] -!- mistersmee has parted #archlinux32
[15:18:58] <T`aZ> yeah, well, the world is not perfect ^^
[15:25:45] -!- mistersmee has joined #archlinux32
[15:55:08] -!- mistersmee has parted #archlinux32
[15:57:33] -!- mistersmee has joined #archlinux32
[15:58:44] -!- mistersmee has parted #archlinux32
[15:59:31] -!- mistersmee has joined #archlinux32
[15:59:41] -!- mistersmee has quit [Client Quit]
[16:11:14] -!- abaumann has quit [Quit: leaving]
[16:20:16] -!- drathir_tor has quit [Ping timeout: 252 seconds]
[16:40:31] -!- mistersmee has joined #archlinux32
[16:42:00] -!- mistersmee has quit [Client Quit]
[16:46:04] -!- mistersmee has joined #archlinux32
[16:50:22] -!- mistersmee has quit [Client Quit]
[17:02:35] -!- drathir_tor has joined #archlinux32
[17:40:14] -!- buildmaster has quit [Read error: Connection reset by peer]
[17:43:06] -!- buildmaster has joined #archlinux32
[20:38:25] -!- Alina-malina has joined #archlinux32