| 27 Jul 2025 |
 aloisw | Dynamic is enough, things break due to linker flag reordering and I want to know whether it happens before pkg-config or after. | 17:51:42 |
| aloisw | (I assume that dynamic is the default on Darwin too, otherwise static would be better.) | 17:52:16 |
 emily | yes. building | 17:54:08 |
| emily | aloisw:
prefix=/nix/store/mr5kgc8kyxba79sdmvc69frq8vk8lrx1-lix-2.94.0-devpre20250727_8bae8f8
libdir=/nix/store/mr5kgc8kyxba79sdmvc69frq8vk8lrx1-lix-2.94.0-devpre20250727_8bae8f8/lib
includedir=/nix/store/xnxhxpwfjq2agwkhi7b24016c27lwj30-lix-2.94.0-devpre20250727_8bae8f8-dev/include
Name: Lix libstore
Description: Lix Package Manager (libstore)
Version: 2.94.0-devpre20250727_8bae8f8
Requires: lix-base lix-util
Requires.private: aws-cpp-sdk-core aws-cpp-sdk-s3 aws-cpp-std-transfer capnp-rpc libcurl libarchive libseccomp libsodium sqlite3
Libs: -L${libdir} -Wl,-force_load -llixstore
| 18:32:35 |
| emily | let me know if you need anything else | 18:32:37 |
| aloisw | Thank you very much, I think I see what's going on now. Meson seems to translate the -l arguments it gets from pkg-config into the library names itself, and then omit duplicates. | 18:35:16 |
| aloisw | Of course when the -l was prefixed with -Wl,-force_load it will not remove the latter, meaning it will apply to some random garbage instead (like other libraries, -Wl,-force_load from other libraries, or the end of the command line). | 18:36:09 |
| emily | when I first used Meson I was like oh this is a pretty good build system | 18:38:12 |
| emily | and then a couple days passed | 18:38:15 |
| aloisw | OF COURSE: ERROR: clang does not know how to do prelinking. | 18:44:10 |
| aloisw | So divergence across platforms, patching meson, or ld -r it is. | 18:53:54 |
 Sergei Zimmerman (xokdvium) | In reply to @emilazy:matrix.org
well it's on the physical package. maybe someone has been wild enough to try and replace just the memory https://forums.macrumors.com/threads/m1-m2-m3-macs-user-memory-upgrades-by-soldering-blocked-by-apple.2410451/ | 19:05:11 |
| emily | the RAM one looks like an Intel Mac | 19:07:05 |
| emily | but yeah I was surprised to learn people have managed it with the NAND | 19:07:06 |
 raitobezarius | In reply to @aloisw:julia0815.de
Can someone with Darwin access then please build https://gerrit.lix.systems/c/lix/+/3778 and post one of the lix-util, lix-expr, lix-fetchers, lix-store pkg-config files? we should maybe get you a Darwin remote builder if you're interested? | 20:00:39 |
 EsperLily [she/her] | oh hey you just merged 3765 as I was reading through it. I was going to ask, why are we still copying the file when we have no chroot? don't we just need to set NIX_SSL_CERT_FILE to point at the caFile in that case? | 23:17:15 |
| EsperLily [she/her] | aaaand i just saw a bug in the new code too, so i just put a comment on it | 23:26:40 |
| 28 Jul 2025 |
 jade_ | ive also noticed this myself. i can possibly throw a no-SIP vm at the problem some day at work. what i would like to learn is what operation is slow in the derivation builder. need a consistent reasonably demanding and reasonably consistent benchmark for this. maybe a build of lix itself, idk. | 00:08:53 |
| emily | my top suspect is QoS priority inversion which a quick taskpolicy call with a non-daemon build should give at least some kind of results for | 00:09:40 |
| emily | even just comparing daemon vs. non-daemon would probably help | 00:09:44 |
| jade_ | mmmm | 00:10:12 |
| jade_ | yeah my test hypothesis is to try to do the genericBuild thing for lix derivation outside the daemon and compare | 00:10:48 |
| jade_ | I concur on your suspicion on QoS being involved, since Kate actually also mentioned this a year ago | 00:11:09 |
| jade_ | * I concur on your suspicion on QoS being involved, since Kate actually also mentioned this a year ago iirc | 00:11:14 |
| emily | if it is priority inversion with the daemon then a quick fix may be as simple as ensuring the daemon re-execs itself with the posix_spawn API to set utility QoS class | 00:12:37 |
| emily | I also suspect something about logging could be very slow | 00:12:53 |
| jade_ | is there a way to get the task policy of a job? | 00:13:01 |
| jade_ | this is DEFINITELY true | 00:13:08 |
| jade_ | the logging in lix is blocking on the daemon | 00:13:15 |
| jade_ | which is totally fucked and we can't really fix it in the current protocol, not easily at least | 00:13:33 |
