| 20 Apr 2022 |
 Janne Heß | In reply to @hexa:lossy.network
and with notifications in intervals This may be doable from the pipeline | 15:52:03 |
| Janne Heß | In reply to @hexa:lossy.network
tracking and notifying how long stuff has been broken This not so much unless graham allows me to scrape more | 15:52:27 |
 hexa | this likely needs a rewrite in something sane as well 😛 | 15:52:50 |
| Janne Heß | Watch me | 15:53:06 |
| hexa | no, not perl | 15:53:17 |
| hexa | bash -> perl is not improvement | 15:53:26 |
| Janne Heß | C99 maybe | 15:55:19 |
 Linux Hackerman | I think you typoed C89 | 15:55:39 |
| hexa | might as well do it in PHP then | 15:56:31 |
|  rh joined the room. | 19:00:08 |
| 21 Apr 2022 |
|  An exploring bot joined the room. | 00:49:36 |
| An exploring bot left the room. | 00:49:37 |
 Vladimír Čunát | Perhaps worth considering to drop i686 ISO from the channel since 22.05? It's not the first time when some dependency got broken and people tend to be reluctant about fixing i686 builds. https://github.com/NixOS/nixpkgs/pull/169281#issuecomment-1104468198 | 04:30:28 |
 tomberek | Some unstable floating point calculation?
The difference between step.newTokenBucketRate and retryTokenBucket.T_GetFillRate() is 0.0019530492986188364, which exceeds EPSILON, where
step.newTokenBucketRate evaluates to 3.452226,
retryTokenBucket.T_GetFillRate() evaluates to 3.4502729507013812, and
EPSILON evaluates to 9.9999999999999995e-07.
i'm hesitant to drop i686 because of a failing test like this
| 05:18:12 |
 Mic92 | Do we know how many users this target has? | 06:10:53 |
| Mic92 | I would expect that even arm64 exceed this by far. | 06:11:08 |
| Mic92 | Does anyone know anyone? | 06:11:22 |
|  arno joined the room. | 06:59:00 |
| 22 Apr 2022 |
| Vladimír Čunát | Maybe CDN could have stats on the i686 ISO downloads? | 06:41:02 |
|  Linux Hackerman is moving: @linus:schreibt.jetzt changed their display name from Linux Hackerman to Linux Hackerman is moving: @linus:schreibt.jetzt. | 07:37:30 |
| Vladimír Čunát | Locally I was getting errors earlier during the build, as some of the compiler/linker steps apparently shot over the 4G RAM. Maybe higher --cores or $(nproc) was causing it. | 08:08:33 |
| Linux Hackerman is moving: @linus:schreibt.jetzt left the room. | 07:40:10 |
| Vladimír Čunát | * Locally I was getting errors earlier during the build, as some of the compiler/linker steps apparently shot over the 4G memory limit. Maybe higher --cores or $(nproc) was causing it. | 08:08:49 |
| Vladimír Čunát | * Locally I was getting errors earlier during the build, as some of the compiler/linker steps apparently shot over 32-bit memory limits. Maybe higher --cores or $(nproc) was causing it. | 08:09:09 |
| Vladimír Čunát | Otherwise I'd have just disabled the test(s) on i686. | 08:09:59 |
 Artturin | In reply to @mic92:nixos.dev
Do we know how many users this target has? We could make a gh issue and Pin it | 11:43:38 |
| 23 Apr 2022 |
|  Amit Levy joined the room. | 03:05:44 |
| Vladimír Čunát | So I hacked around it for now (by platform-specific downgrade), and thus it's perhaps easy to keep it for 22.05, but it might be worth considering to drop after forking off (and thus for > 22.05). | 10:53:22 |
| Vladimír Čunát | BTW, having nixos.tests.zfs.installer.i686-linux in channel-blocking set seems really absurd. | 10:55:58 |
| 25 Apr 2022 |
| Mic92 | Agreed. zfs is not really suited for 32-bit let alone legacy x86 | 07:16:57 |
