| 13 Aug 2025 |
 emily | but from my previous looking, getting rid of Qt 5 would be quite a lot yes | 16:24:33 |
| emily | hard to say how much post-compression | 16:24:38 |
| emily | K900: Maliit is bringing in some ~110 MiB of Qt 5 yeah | 16:42:12 |
| emily | (uncompressed though) | 16:42:15 |
 K900 | Oh wow I expected worse | 16:42:43 |
| emily | let me look at what the worst remaining ISO closure offenders are | 16:43:41 |
| emily | so I believe we have the firmware in there like at least twice for the two initrds | 16:56:52 |
| emily | we should probably split that out into a separate initrd and pass both or something | 16:56:59 |
| emily | I forget whether you can do that without a stub though | 16:57:07 |
| emily | I guess we can because microcode? | 16:57:24 |
| emily | but not sure how big a lift it'd be | 16:57:37 |
| emily | we have two Pythons and I don't know why | 16:58:35 |
 Grimmauld (any/all) | that should be fixable... | 16:59:06 |
| emily | we have mesa-demos | 16:59:25 |
| emily | I assume for glxgears | 16:59:27 |
| emily | not sure how much it pulls in beyond that | 16:59:33 |
| emily | but it's surprisingly big | 16:59:48 |
| Grimmauld (any/all) | i need to make a notes file, i keep forgetting how to build the official iso... | 17:00:36 |
| Grimmauld (any/all) | ah shell history still had it :) | 17:01:15 |
| Grimmauld (any/all) | time to poke why there is multiple pythons | 17:01:26 |
| emily | K900: okay so Mesa's closure is 941 MiB of which 540 MiB is LLVM. 161 MiB of that is the actual libLLVM.so.19.1 that I presume we can't do without | 17:02:59 |
| emily | and almost all of the rest is | 17:03:05 |
| emily | static libraries | 17:03:08 |
| emily | in the lib output | 17:03:10 |
| emily | because lol | 17:03:12 |
| Grimmauld (any/all) | that sounds silly | 17:03:29 |
| emily | probably if we can move those to dev we can chop off like a hundred megs of compressed ISO or more easy | 17:03:37 |
| emily | not just from the ISO but from like… every NixOS system? | 17:03:56 |
| emily | how has nobody noticed this before | 17:04:00 |
| emily | also is Mesa really meant to have runtime deps on Python and elfutils | 17:04:28 |
