Gabriel Scherer
Gabriel Scherer
We have an OCAMLRUNPARAM parameter to set the OCaml stack size (it is also controllable through `Gc.set`). I would find it reasonable to also have a parameter to control the...
I asked about this at the "runtime meeting" today, and @stedolan mentioned that Jane Street has a patch to use *multi-letter* setting names for OCAMLRUNPARAM, which sounds pleasantly usable when...
I still think that the most natural behavior would be for newly spawned thread to inherit their parent's setting. (Note: this may not matter much in many uses of a...
I am not familiar with nbdkit and the behavior of its `-s` option, so I could not understand the comment. (It also has an unclosed parenthesis I think.) Would you...
I find this discussion a bit draining, and I am worried by the potential that it would turn into a flamewar with too many people giving their personal opinion on...
I could reproduce both under 5.4.0. The second example is a clear user-interface bug and I believe that a fix is not too hard -- it's a matter of correctly...
I added `is_val` as requested. I also realized that I had forgotten to push my tests (the usual "forgotten `git add`" issue), so this is also done now.
It looks like your computer suddenly starts returning `2048` on certain reads from memory. It might be a memory error, could you run tests on your memory?
(The testsuite is failing and @damiendoligez needs to look at it.)
In the gc-pacing.pdf document: > `s` is the amount of GC work done per word allocated in the Sweep phase I read this as "amount of GC work done per...