| 1 Jun 2021 |
 Regnat | In reply to @rgrunbla:matrix.org
hey Regnat , the CA experimental features actually fix the compilation, but it does not really work when you use it as the input for another flakes, e.g https://0x0.st/-2Ic.txt Yes, as I said, this isn’t really supposed to work 🙃 | 07:24:53 |
 Reventlov | Redacted or Malformed Event | 07:25:02 |
| Regnat | I’m not sure why you need hello to be a fixed-output derivation tbh | 07:25:31 |
| Reventlov | Redacted or Malformed Event | 07:25:59 |
| Reventlov | Redacted or Malformed Event | 07:26:47 |
| Regnat | Ugh that’s ugly ☹️ | 08:27:20 |
| Regnat | If all you want is to have all the submodules fetched, maybe you can remove the build artifacts that have a self-ref in the postBuild | 08:28:02 |
| Reventlov | Redacted or Malformed Event | 08:32:17 |
| Reventlov | Redacted or Malformed Event | 08:32:35 |
| Reventlov | Redacted or Malformed Event | 08:33:00 |
| Reventlov | Redacted or Malformed Event | 08:33:31 |
 Alyssa Ross | @eelco:nixos.org do you know why i*86 platforms are forced to i686 for the Nix system in configure.ac? | 19:26:04 |
| Alyssa Ross | in #exotic:nixos.org we're wondering what the way to go would be for somebody to build e.g. i486 packages | 19:26:43 |
 niksnut | Dunno, it has been that way for a long time. Probably to avoid getting i786 or whatever. | 20:11:10 |
| niksnut | Oh, here is the explanation:
commit 966bd9d19fb43b6a6022572eb7d80c1e03bfaf3e
Author: Eelco Dolstra <e.dolstra@tudelft.nl>
Date: Fri Aug 13 09:57:51 2004 +0000
* WTF? More canonical system name problems ("athlon-linux" instead of
"i686-linux").
| 20:12:38 |
| Alyssa Ross | ah great, thanks | 20:19:29 |
| Alyssa Ross | I only managed to find the commit from Ludo in 2011 that just said "it's always been this way" | 20:19:42 |
| 2 Jun 2021 |
 Jan Tojnar | Did something change regarding the cycle detection? Now I am getting “error: cycle detected in build of … in the references of output 'lib' from output 'jack'” and the output paths for the cyclic outputs are not in the store, not even with --keep-failed | 01:38:50 |
| Jan Tojnar | with stable Nix, the output paths are listed in the error message (so I do not have to fish for them in the build log) and they exist even without --keep-failed | 01:40:46 |
| Alyssa Ross | niksnut: while you're merging Nix PRs, any thoughts on https://github.com/NixOS/nix/pull/4773? :) | 11:35:55 |
| niksnut | Does that work with lowdown? | 11:36:49 |
 sterni | niksnut: https://kristaps.bsd.lv/lowdown/lowdown.3.html#LOWDOWN_DEFLIST | 11:37:32 |
| sterni | seems like you've already turned it on in libcmd/markdown.cc | 11:38:24 |
| Alyssa Ross | oh, do they end up in a man page? | 11:39:32 |
| sterni | Alyssa Ross: :doc on the repl is rendered with lowdown's term rendering target | 11:40:27 |
| Alyssa Ross | ahh | 11:41:38 |
| Alyssa Ross | let me test | 11:41:39 |
| sterni | we are in a bit of an unfortunate situation where we render the manuals with pandoc atm, ryantm is implementing the potential next manual renderer using cmark and nix is using lowdown for rendering markdown on the terminal | 11:41:52 |
| sterni | cmark is extensible, so lowdown is probably the constraining factor in the equation | 11:42:45 |
| Alyssa Ross | ugh, I don't have the build cached any more | 11:43:00 |
