3 Apr 2025 |
Winter | no recursive nix required | 22:01:33 |
Winter | apparently i am stupid and for now you technically need it to do some stuff, but ideally once something exists to replace that functionality (basically just adding many drvs from one derivation build), then it’ll be useful | 22:10:19 |
jade_ | yes, it ultimately requires a limited replacement for the daemon protocol that is less fucked up and thus could be stabilized as derivation ABI | 22:26:18 |
jade_ | fortunately there is agreement between cppnix and us about this | 22:26:34 |
jade_ | so they are not going to surprise stabilize it without a clear answer to that issue | 22:26:47 |
| dantefromhell changed their display name from dantefromhell to MATRIX dantefromhell. | 22:34:31 |
| dantefromhell changed their display name from MATRIX dantefromhell to dantefromhell. | 22:36:24 |
4 Apr 2025 |
John Ericson | If you have a good batching protocol | 00:15:00 |
John Ericson | You don't even need a reply I think | 00:15:00 |
John Ericson | You can submit all store objects and derivations you need, in a very simple "language" to substitute some variables | 00:15:48 |
John Ericson | And you don't need a response even because the builder doesn't need to do anything more | 00:16:34 |
John Ericson | The builder can just be killed, and derivation succeed/fail based on whether the request is valid | 00:17:24 |
John Ericson | CC @winter:catgirl.cloud | 00:17:37 |
jade_ | https://github.com/curl/curl/issues/16955 AAAAAAAAAAAAAAAAAAAA okay we have done it | 00:28:57 |
jade_ | raitobezarius: would you be able to review https://gerrit.lix.systems/c/lix/+/2887?usp=search and https://gerrit.lix.systems/c/lix/+/2931?usp=search when you're free? | 00:42:55 |
raitobezarius | Will do, thx for the ping | 00:44:31 |
just1602 | In reply to @jade_:matrix.org https://github.com/curl/curl/issues/16955 AAAAAAAAAAAAAAAAAAAA okay we have done it The quality of this bug report !!!!! | 01:04:29 |
jade_ | i try!!! | 01:04:35 |
just1602 | In reply to @jade_:matrix.org i try!!! And succeeded | 01:05:16 |
jade_ | seems like the reason that lix 2.93 is having build interrupt issues on macos is that previous versions were only working by accident >:( | 01:06:54 |
jade_ | namely it seems that MonitorFdHup doesn't actually work, which is surprising | 01:07:17 |
jade_ | previous versions threw an error while writing to a dead socket, rather than getting interrupted by the socket being hung up | 01:09:00 |
jade_ | although, well, maybe it is not that surprising that MonitorFdHup doesn't work given the number of comments of um. oral history. | 01:09:33 |
jade_ | anyway i clearly need to write a damn test | 01:09:42 |
uep | it's outstanding | 01:20:12 |
jade_ | did i think that the macOS kernel would be so easy as to be broken in an easily reproducible way? well, my naïevété shows through haha https://gerrit.lix.systems/c/lix/+/2933 | 01:47:19 |
jade_ | * did i think that the macOS kernel would be so easy as to be broken in an easily reproducible way? well, my naïvété shows through haha https://gerrit.lix.systems/c/lix/+/2933 | 01:47:26 |
jade_ | ...... okay i am going to log off. i am flummoxed with this bug. it is fine with socat, but not with a real nix | 01:55:47 |
jade_ | i wonder if we have found a kernel bug | 01:55:53 |
jade_ | beats me | 01:55:57 |