| 8 Dec 2025 |
 emily | but, yes, it would actually be best to fix this upstream then | 17:52:06 |
| emily | since their version is just useless | 17:52:10 |
 Ihar Hrachyshka | I will update the issue they have later with the above, really cool stuff | 17:52:24 |
| emily | I suggest replacing LLAMA_INSTALL_VERSION with LLAMA_BUILD_NUMBER upstream | 17:52:26 |
| emily | that's a two line diff | 17:52:34 |
| Ihar Hrachyshka | would take me a lot longer, appreciate it | 17:52:37 |
| emily | and in fact we could just vendor this patch locally while waiting for upstream | 17:52:47 |
| emily | only infelicity is that if they ever do actual versions the ordering will be weird | 17:52:58 |
| emily | but I doubt anything actually cares about the value | 17:53:09 |
| emily | that will start failing when they go over 65536 commits too fwiw | 17:53:49 |
| emily | or well, start truncating | 17:53:54 |
| emily | I'm not convinced anything actually cares about the "current version" field of a library though | 17:54:29 |
 Randy Eckenrode | Many times projects just set junk versions. The versions for llbuild’s framework are 1 and 1. | 17:54:31 |
| emily | yeah | 17:54:48 |
