Landlocked redo?

[Joseph Garvin]

Build isolation is one of the things I miss from more "enterprise" systems like Bazel, but apparently the new Linux landlock APIs are sufficient for implementing it in a way that's easy to integrate into existing tools:

https://justine.lol/make/?s=09

[Charles Eckman]

Thanks for sharing, this is a neat post!

(My understanding is-) the biggest philosophical / structural difference between makes / make-likes (incl. Bazel) and redo is that redo *does not* specify the dependency graph up front. So- I'm not sure if / how "strict dependency checking" would apply to redo without a philosophical shift?

Output isolation might be useful- 

"RWC the output file and a temporary directory, RX everything else" as the policy? I'm not sure how that would play with the "rebuild directory as a side effect" pattern- maybe there are other decent ways of producing "an output directory" that would allow this isolation?

On Sun, Aug 7, 2022, 8:47 AM Joseph Garvin <k04...@gmail.com> wrote:

Build isolation is one of the things I miss from more "enterprise" systems like Bazel, but apparently the new Linux landlock APIs are sufficient for implementing it in a way that's easy to integrate into existing tools:

https://justine.lol/make/?s=09

--
You received this message because you are subscribed to the Google Groups "redo" group.
To unsubscribe from this group and stop receiving emails from it, send an email to redo-list+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/redo-list/CAH%3DMJ4Y%3DONd9bqN5DHS6jS_4vgtfzjfrhSAMGs2NnaeSenGqfg%40mail.gmail.com.

[Joseph Garvin]

I think you your do scripts when executing would be restricted to only reading files that have earlier been redo-ifchange'd in the script. And yeah restrict writes to the output.