http_open/3 hangs after being called twice

[thierrymarianne]

Hello 👋🏼,

I've noticed this related issue #2470.
Since it was not closed in between, and I've encountered a similar issue,

I've added the same test case to two separate branches,
and the issue is reproducible for both rebis-dev brand and v0.9.4 tag.

Pull-request #2833 was opened by mistake (when I intended to trigger the ci workflow from my fork,
sorry about that). However, it contains a module which can be loaded
to have a closer look at the issue.

By debugging, one can see that execution gets stuck on third request being sent at
https://github.com/mthom/scryer-prolog/blob/rebis-dev/src/machine/system_calls.rs#L4497

With some guidance, I could perhaps offer a fix if someone sees how to proceed further,
which may or may not help in closing #2470 as well.

[thierrymarianne]

It seems that replacing futures::executor:block_on with tokio::block_in_place can help in identifying the actual issues in the way how the Ok(resp) branch is handled.

By removing all resource_error_call_result! macros,
direct unwrapping leads to a solution (no more hanging calls to http_open/3).

Of course, it would be better to fix the macro (in rebis-dev),
but I don't know how it should be done.

[thierrymarianne]

A tentative pull-request to check from ci workflows that nothing broke from the futures::executor:block_on replacement with tokio::block_in_place appears to be ok.
https://github.com/thierrymarianne/contrib-scryer-prolog/actions/runs/13529411529/job/37807772388