Mark Shinwell

Results 10 issues of Mark Shinwell

As has been thought about

At link time record the paths used to locate various compilation artifacts; these could then be used to augment the .cmt search path in the debugger. This should mean that...

It shouldn't just say "x = 0" when there is a better error to be printed.

Set breakpoint on Flambda_to_clambda.to_clambda Look at the "env" argument. The value has six fields but the extracted type has 14. Six is correct for Flambda_to_clambda.Env.t. It looks like it's found...

- Boxed values of polymorphic variant type are currently not printed at all. - @lpw25 claims that the existing code isn't correct in any case.

The `caml_domain_external_interrupt_hook` strikes again. When `caml_set_minor_heap_size` is called, the minor heap is emptied by calling `caml_minor_collection`, before the existing minor heap for the domain concerned is freed and reallocated. Except...

bug
runtime-system

Here is a nice straightforward one for the New Year! In [this PR on the Flambda 2 repo](https://github.com/ocaml-flambda/flambda-backend/pull/2204), which is 5.1.1 plus a few backports, I describe a bug where...

bug
runtime-system

Please see description in `rfcs/symbols.md`. ([Rendered](https://github.com/ocaml/RFCs/blob/225743e9e55616268513753c744009c71a738965/rfcs/symbols.md))

rfc
backend

As [described in more detail here](https://github.com/ocaml-flambda/flambda-backend/pull/3960), the Cmm machtypes for "pointer + offset" calculations used for the various string read/write operations appear to be wrong, which can lead to the...

bug

This is an edited version of the AI text in https://github.com/ocaml/ocaml/pull/14052, together with moving it into a new file `AI.md`.

no-change-entry-needed
documentation
CI: Skip testsuite