| 6 Apr 2026 |
 mio | pentane: how about add it to robotnix grapheneos's flash_all.sh too to align with upstream behaviour? | 12:43:40 |
| mio | * pentane: how about adding it to robotnix grapheneos's flash_all.sh too to align with upstream behaviour? | 12:43:46 |
| mio | I did attempt to do it on grapheneos and lineageos. It works for grapheneos. For now I don't have a lineageos device with gki kernel to test prepatched kernelsu kernel on lineageos | 12:45:20 |
| mio | I really need incremental builds. I change one line in one nix file and wait for two hours for robotnix to build. Then repeat | 12:50:59 |
 pentane | there is an open pr for that, but it stalled after i requested changes. Might merge it and implement the changes by myself | 12:57:17 |
| pentane | welcome to my world 💀 | 12:57:26 |
| pentane | i recommend building android in a devshell (config.build.debugEnterEnv) if you wanna iterate | 12:58:13 |
| mio | Redacted or Malformed Event | 13:02:23 |
| mio | Redacted or Malformed Event | 13:04:47 |
| pentane | Brief reminder that LLMs are off-topic here | 13:05:13 |
| pentane | * Brief reminder to stay on-topic here | 13:05:35 |
|  esch joined the room. | 14:53:27 |
| 8 Apr 2026 |
 atemu12 | Not sure whether I published it but I recently had the idea to pierce the sandbox for the gradle cache for the java stuff that can't be ccached. Though now that I've written it out I realise that it might not even use gradle and calls javac directly, hm | 02:03:37 |
| atemu12 | Mozilla has a caching tool for rust though that ought to be usable in the sandbox somehow | 02:04:29 |
 jack | I remember coming across a nix project that did this for gradle. Lemme dig around a bit. | 04:09:56 |
 Sophia | I mean the robotnix project does offer a ccache. Though for me it feels like it doesn't do much | 05:56:13 |
| atemu12 | ccache only caches the c/c++ bits | 07:24:33 |
| atemu12 | A lot of the AOSP is Java and some rust is in there too | 07:25:18 |
| 10 Apr 2026 |
 neobrain | https://nlnet.nl/project/Shinobi/
Are any of you behind this or is that a separate effort? | 06:45:19 |
| pentane | @jaen:matrix.org was this your grant? | 12:35:13 |
 jaen | Yes, that's the one | 12:36:52 |
| jaen | * | 12:38:33 |
| neobrain | nice, congrats! | 13:13:04 |
 magic_rb | Lol, you found out you got the grant cause you got pinged here? | 13:21:22 |
| magic_rb | Peak | 13:21:23 |
| jaen | It would be funny if I did, but no - I actually do read e-mails from time to time xD | 15:23:40 |
| jaen | I guess now I have to actually go and build this thing for real. | 15:26:18 |
| jaen | Gotta hold Ericson by his "we're making dyndrvs generally usable this year" because when I transitioned my PoCs from parsing at scale to building at scale it turns there sure is a bit of overhead to solve xD | 15:26:28 |
| neobrain | To me as a layperson the project sounds like it's solving a similar problem space as soongnix did.
Are they related at all or am I thinking in the wrong direction? | 19:50:21 |
| atemu12 | Depends on which way you build it. If you take the simple approach of basically doing nix-ninja, you don't interact with soong at all but that's unlikely to work in practice because drvs don't scale into the 100000s. | 23:05:53 |
