| 25 Apr 2022 |
 etu | Just as we should drop php74 and change to php81 as default by then | 08:04:39 |
 Pol | Oki, could you comment on the thread, so this information is not lost in the matrix's void :) | 08:05:03 |
 Andreas Schrägle | In reply to @etu:semi.social
Just as we should drop php74 and change to php81 as default by then You can add that to the 22.11 blockers project | 08:08:10 |
| etu | Oh, great suggestion :) | 08:11:47 |
| etu | Done! | 08:13:30 |
| Pol | How do you do that? | 08:18:15 |
| 26 Apr 2022 |
| Pol | https://dpaste.org/a3euL | 05:23:01 |
| Pol | Looks like the GCC update broke PHP < 7.4 compilation | 05:23:13 |
| Andreas Schrägle | the easiest way around that is probably just using gcc < 11 | 07:22:54 |
| Andreas Schrägle | Also, how does nix-php (is that what it's called) handle openssl_1_0_2 finally not being a thing anymore in nixpkgs? | 07:23:27 |
| Pol | It's using a patch | 08:33:54 |
| Pol | In reply to @andreas.schraegle:helsinki-systems.de
the easiest way around that is probably just using gcc < 11 How to specify the gcc version to use then? | 08:34:06 |
| Andreas Schrägle | There's a lot of ways, but the easiest is probably using gcc10Stdenv instead of stdenv.
Although that probably breaks the build on aarch64, if that works right now, for reasons. | 08:39:31 |
| Pol | Oh... damn :S | 08:40:10 |
| Pol | Jan Tojnar: WDYT ? ^^ | 08:40:21 |
| Andreas Schrägle | In reply to @andreas.schraegle:helsinki-systems.de
There's a lot of ways, but the easiest is probably using gcc10Stdenv instead of stdenv.
Although that probably breaks the build on aarch64, if that works right now, for reasons. I can link the reasons, but basically it's because libgcc_s.so is from the bootstrap tarball. Or something like that. It's complicated -.- | 08:43:08 |
| Pol | Indeed, this is way too complex for me. | 08:45:01 |
 Jan Tojnar | Pol: I would look for a patch that allows it to build with newer gcc | 09:59:25 |
| Pol | I already checked, but couldn't find anything. Nobody's looking to compile PHP < 7.4 with newest GCC ! | 10:04:12 |
| Jan Tojnar | well, it must have been fixed in PHP 8 at some point | 10:28:02 |
| Jan Tojnar | so perhaps there is patch in PHP repo that we can backport | 10:28:17 |
| Jan Tojnar | if we make nix-phgs able to build from git, we could try bisecting it | 10:29:28 |
| Andreas Schrägle | In reply to @jtojnar:matrix.org
if we make nix-phgs able to build from git, we could try bisecting it why isn't it, right now? I remember a pr that made the php in nixpkgs build from git, which we probably didn't merge for reasons. maybe it was even by Pol? or by Izorkin? maybe I'm making all of this up. well never find out, unless someone searches github. | 16:43:57 |
| Pol | In reply to @andreas.schraegle:helsinki-systems.de
why isn't it, right now? I remember a pr that made the php in nixpkgs build from git, which we probably didn't merge for reasons. maybe it was even by Pol? or by Izorkin? maybe I'm making all of this up. well never find out, unless someone searches github. I tried indeed, but I think I never submitted a PR | 16:44:26 |
| Jan Tojnar | In reply to @drupol:matrix.org
I tried indeed, but I think I never submitted a PR you did https://github.com/NixOS/nixpkgs/pull/155012 | 18:07:14 |
| Jan Tojnar | In reply to @andreas.schraegle:helsinki-systems.de
why isn't it, right now? I remember a pr that made the php in nixpkgs build from git, which we probably didn't merge for reasons. maybe it was even by Pol? or by Izorkin? maybe I'm making all of this up. well never find out, unless someone searches github. we usually prefer the official distribution method to reduce the chance of divergences. For PHP it is tarballs. | 18:09:50 |
| Pol | In reply to @jtojnar:matrix.org
you did https://github.com/NixOS/nixpkgs/pull/155012 Ah that controversial one ! :) | 18:19:34 |
| Pol | https://github.com/NixOS/nixpkgs/commit/52f8cf58a4504e5e219faebffa51033e400e3aec#r72228075 | 18:29:22 |
| Jan Tojnar | Pol: I added the master version here https://github.com/fossar/nix-phps/pull/104 | 19:15:52 |
| Jan Tojnar | but currently it fails to build | 19:16:03 |
