Post #4170142
2026-07-28 17:54 UTC
RE: https://social.v.st/@wfk/116958566117813410
I've been thinking a lot about the whole LLM based analysis of open source projects. Such as the whole 400+ CVEs in the Linux kernel recently found. For years as codebase bloat has become the norm for most projects, it's become standard to go with the second option of this quote. So complicated there are no obvious deficiencies.
Given how huge many projects have not got, perhaps it's time to rethink our approach.
1/n
Replies (1)
-
@quixoticgeek@social.v.st 2026-07-28 17:57
The chrome browser is in the order of 40m lines of code (LoC). Systemd is ~2m LoC. There is no way with code bases this big that one person can be on top of all of it, to know what's going on, to understand it all. It's so complex there are no obvious deficiencies. Maybe if we had more projects that were simple, so simple there are obviously no deficiencies, the LLM tools being used to find security flaws would have a much harder job to do. 2/n