Sender | Message | Time |
---|---|---|
17 Jul 2022 | ||
Redacted or Malformed Event | 02:23:20 | |
*
https://github.com/ModelioOpenSource/Modelio/blob/v5.1.0/doc/plugins/en/documentation.copyright.org/build.properties
| 02:43:18 | |
having to download all dependencies everytime i need to build the derivation is probably the worst developer experience i have had on nixpkgs | 03:56:37 | |
19 Jul 2022 | ||
hi! | 16:31:31 | |
fzakaria: hi there what's up? that nix/jvm was pretty good btw. is it really a lost cause trying to integrate nix and maven, gradle? what do you think about current ant state in nix in comparison to maven and gradle? | 16:56:45 | |
* fzakaria: hi there what's up? that nix/jvm talk was pretty good btw. is it really a lost cause trying to integrate nix and maven, gradle? what do you think about current ant state in nix in comparison to maven and gradle? | 16:58:08 | |
I think gradle is a lot easier since it can generate a lock file | 16:59:16 | |
Although that lock file might be missing everything; it tends to only include runtime dependencies and not preprocessors, plugins etc.. | 16:59:35 | |
If we could hack Maven to legitimately output all dependencies downloaded in a fresh m2 store then its doable | 16:59:36 | |
noChRoot is also pretty doable but not "pure nix" | 17:00:01 | |
In reply to @fzakaria:one.ems.hostis there a pattern for which dependencies aren't put in the m2 store? like maybe plugins are left out? | 17:11:08 | |
well if you write a plugin to analyze dependencies you cannot get those | 17:19:36 | |
so you need to hack maven deeper | 17:19:44 | |
the other problem i had was that an artifact can be vended from multiple artifactories | 17:20:09 | |
same version but different SHA :( | 17:20:13 | |
which is silly | 17:20:22 | |
22 Jul 2022 | ||
In reply to @fzakaria:one.ems.hosti've encountered it while packaging this https://github.com/NixOS/nixpkgs/pull/154188/checks?check_run_id=7473713573 using the usual fod. so basically non-reproducible | 18:47:02 | |
In reply to @fzakaria:one.ems.host* i've encountered maybe a similar issue while packaging this https://github.com/NixOS/nixpkgs/pull/154188/checks?check_run_id=7473713573 using the usual fod. so basically non-reproducible | 18:52:55 | |
24 Jul 2022 | ||
18:35:51 | ||
1 Aug 2022 | ||
09:13:59 | ||
09:14:12 | ||
In reply to @fzakaria:one.ems.hostI think we did this, in a very brutish way, for clojure-nix-locker: https://github.com/bevuta/clojure-nix-locker/blob/master/default.nix#L61 | 12:16:19 | |
we set that thing to make maven use the freshly created directory for its cache, then run a user-provided build command (that hopefully pulls everything needed into the cache), then walk the cache to generate a lockfile | 12:17:06 | |
this works perfectly fine for clojure stuff, but that might be because that tends to make only relatively simple usage of maven. I can see this breaking in the presence of advanced maven features. | 12:18:00 | |
you also need to set $HOME IIRC, we do that in a different place | 12:19:24 | |
6 Aug 2022 | ||
17:42:53 | ||
20:35:45 | ||
13 Aug 2022 | ||
03:15:31 | ||
05:06:01 | ||
17 Aug 2022 | ||
20:23:36 |