| 19 Mar 2022 |
 @elvishjerricco:matrix.org | yea, I was going to add stuff specifically for parsing Exec* options in service files to make sure it was as minimal as possible, rather than just any referenced path | 09:38:14 |
 bobvanderlinden | I see that as an optimization ;) it's good to do so. My solution is crude indeed, but it does get the job done, also for things like initramfs.systemd.services.myservice.script = "... ${somePackage}/bin/something ..."; where ExecStart refers to a bash script, that refers to a bin file. It can go deep I guess 😅 | 09:40:43 |
| @elvishjerricco:matrix.org | we'll have to play with it and see the size consequences | 09:41:35 |
| bobvanderlinden | The initramfs of mine was 30M last time I ran it | 09:41:56 |
| @elvishjerricco:matrix.org | Oof. Last time I was making progress on this stuff I wasn't happy that it was up to 20M | 09:42:17 |
| @elvishjerricco:matrix.org | Old is like 10M | 09:42:25 |
| bobvanderlinden | I thought first get everything working with the reusable initramfs.systemd.services. See what else needs to go in contents (hopefully only the /etc and /bin symlinks). Once that works, see whether the find-dependencies can be optimized more without missing any dependencies | 09:42:50 |
| @elvishjerricco:matrix.org | I think size takes priority over find-dependencies finding everything | 09:43:24 |
| @elvishjerricco:matrix.org | it's not such a big deal if we have to add some manual paths | 09:43:31 |
| bobvanderlinden | Ah! I compared with the original initrd, it seemed those were similar. | 09:43:42 |
| bobvanderlinden | agreed 👍️ | 09:43:51 |
| bobvanderlinden | it's fine to remove the dependency finding in arbitrary files | 09:44:44 |
| @elvishjerricco:matrix.org | We should definitely be getting rid of all the stuff like extraUnits and stuff that I did in favor of reusing the systemd modules though. | 09:46:17 |
| bobvanderlinden | it get a bit messy anyway when we want to support boot.initrd.extraUtilsCommands | 09:46:18 |
| @elvishjerricco:matrix.org | We do not want to support stuff like boot.initrd.*Commands I'm pretty sure | 09:46:37 |
| @elvishjerricco:matrix.org | That's going to be a road that leads only to madness | 09:46:45 |
| @elvishjerricco:matrix.org | Given that this should be an opt-in optional initrd, it should be clean slate | 09:46:56 |
| bobvanderlinden | I think it eases the migration from initrd to initramfs.systemd. Helps keep the PR smallish and easier to merge | 09:47:54 |
| @elvishjerricco:matrix.org | Well | 09:48:01 |
| @elvishjerricco:matrix.org | It should come in stages. | 09:48:10 |
| @elvishjerricco:matrix.org | One PR that just adds bare bones functionality | 09:48:18 |
| @elvishjerricco:matrix.org | and one PR after another restoring old features | 09:48:25 |
| @elvishjerricco:matrix.org | The *Commands things are really really antithetical to the systemd approach in the first place | 09:48:44 |
| @elvishjerricco:matrix.org | and it's just going to make things harder anyway | 09:48:49 |
| bobvanderlinden |
Given that this should be an opt-in optional initrd, it should be clean slate
Ah like that. Merge a clean slate and start adding initramfs.systemd configs to modules that currently use extraUtilsCommands?
| 09:48:52 |
| bobvanderlinden | *
Given that this should be an opt-in optional initrd, it should be clean slate
Ah like that. Merge a clean slate and start adding initramfs.systemd configs to modules that currently use extraUtilsCommands?
| 09:49:04 |
| bobvanderlinden | indeed, alright, that also sounds like a good idea | 09:49:17 |
| @elvishjerricco:matrix.org | It shouldn't be too bad either; initrd doesn't really support that much right now | 09:49:37 |
| @elvishjerricco:matrix.org | The excessive number of LUKS features will probably be the hardest part :P | 09:49:51 |
| bobvanderlinden | at some point deprecate everything from boot.initrd | 09:49:52 |
