| 5 Jul 2026 |
 Diogo Correia | * Not sure how much we should modify vchord without risking introducing bugs, but if that's the only way to not have the package break, sure
Last thing I want to do is to corrupt people's postgres databases | 12:08:03 |
 emily | dropping one ISA extension should be safe | 12:16:46 |
| emily | we can bump LLVM in a cycle or two | 12:16:56 |
| emily | was it said that vchord is unmaintained? | 12:17:09 |
 Lach | pgvector is, vchors is a replacement and maintained | 12:17:34 |
| emily | this is what I did for Swift | 12:17:57 |
| Lach | My solution is a bit dirtier than that, I cfg-out x86_64-v4 support and then make it never hit this path | 12:18:47 |
| emily | scrap the cycle for like 200 packages? | 12:19:00 |
 K900 | Is there really this little Qt on Darwin? | 12:20:46 |
| emily | it might be more but it's not tons and tons | 12:20:59 |
 hexa | asyncssh … you dropped a test disable that still fails here, readding it | 12:21:04 |
| emily | mostly just because most downstream stuff is broken unrelatedly | 12:21:14 |
| emily | like KDE stuff | 12:21:21 |
| emily | anyway | 12:21:29 |
| emily | how about just adding the lid stuff to the qtModule thing first | 12:21:42 |
| emily | and seeing if it still happens with apps or if that's enough | 12:21:52 |
| emily | I don't have the time today but it should be easy to copy the 500 other times it's been done this cycle | 12:22:13 |
| K900 | I ran it like three times and it didn't fail | 12:22:42 |
| K900 | I guess it's also somehow setup dependent? | 12:22:49 |
| hexa | could be | 12:22:53 |
| hexa | zfs? | 12:23:02 |
| emily | let's just land that then | 12:24:39 |
| emily | sounds noninvasive | 12:24:43 |
| K900 | No | 12:28:01 |
| hexa | then that may very well be the issue | 12:28:16 |
| emily | @K900 do you feel like testing this for both Qt 5 and 6? if you grep cmake.*LLD you'll find what to copy. Qt 6 qtdeclarative blocks 585 packages on Darwin and Qt 5 blocks the channel. if a downstream app builds then the hack can probably stay in Qt without needing to propagate it out | 13:11:52 |
| emily | I'd also suggest someone just drop GIMP 2 from the channel blockers. I know it's probably there as a proxy for GTK 2 but I'm fine if GTK 2 breaks on Darwin. | 13:12:32 |
| emily | oops @k900:0upti.me. thanks Element X | 13:13:35 |
| K900 | In reply to @emilazy:matrix.org
@K900 do you feel like testing this for both Qt 5 and 6? if you grep cmake.*LLD you'll find what to copy. Qt 6 qtdeclarative blocks 585 packages on Darwin and Qt 5 blocks the channel. if a downstream app builds then the hack can probably stay in Qt without needing to propagate it out Sorry, migraine | 13:14:42 |
| K900 | I'll probably be out for today | 13:14:47 |
