+ extunix 0.3.1

[ygrek]

follow-up to #18478 with compatibility maintained

[mseri]

Are the test failures on centos-7 and oraclelinux-7 expected?

#         test alias test/runtest (exit 1)
# (cd _build/default/test && ./test.exe)
# ....................F.............................
# ==============================================================================
# Failure: tests:9:pts
# 
# Unix_error : grantpt() : Invalid argument
# ------------------------------------------------------------------------------
# Ran: 40 tests in: 0.04 seconds.
# FAILED: Cases: 40 Tried: 40 Errors: 0 Failures: 1 Skip:  0 Todo: 0 Timeouts: 0.
# .
# ==============================================================================
# Failure: tests:9:pts
# 
# Unix_error : grantpt() : Invalid argument
# ------------------------------------------------------------------------------
# Ran: 40 tests in: 0.04 seconds.
# FAILED: Cases: 40 Tried: 40 Errors: 0 Failures: 1 Skip:  0 Todo: 0 Timeouts: 0.
# ....
# ==============================================================================
# Failure: tests:9:pts
# 
# Unix_error : grantpt() : Invalid argument
# ------------------------------------------------------------------------------
# Ran: 40 tests in: 0.04 seconds.
# FAILED: Cases: 40 Tried: 40 Errors: 0 Failures: 1 Skip:  0 Todo: 0 Timeouts: 0.
# 
# ==============================================================================
# Failure: tests:9:pts
# 
# Unix_error : grantpt() : Invalid argument
# ------------------------------------------------------------------------------
# Ran: 40 tests in: 0.04 seconds.
# FAILED: Cases: 40 Tried: 40 Errors: 0 Failures: 1 Skip:  0 Todo: 0 Timeouts: 0.

And

#         test alias test/runtest (exit 1)
# (cd _build/default/test && ./test.exe)
# ....................F................................
# ==============================================================================
# Failure: tests:9:pts
# 
# Unix_error : grantpt() : Invalid argument
# ------------------------------------------------------------------------------
# Ran: 40 tests in: 0.05 seconds.
# FAILED: Cases: 40 Tried: 40 Errors: 0 Failures: 1 Skip:  0 Todo: 0 Timeouts: 0.
# 
# ==============================================================================
# Failure: tests:9:pts
# 
# Unix_error : grantpt() : Invalid argument
# ------------------------------------------------------------------------------
# Ran: 40 tests in: 0.05 seconds.
# FAILED: Cases: 40 Tried: 40 Errors: 0 Failures: 1 Skip:  0 Todo: 0 Timeouts: 0.

[ygrek]

will need to look into tests failure, but I will consider it not immediately blocking the release

[mseri]

Mmh, it appeared also on another test:

-         test alias test/runtest
- ...........................................
- Ran: 40 tests in: 0.04 seconds.
- OK
- .
- Internal error, please report upstream including the contents of _build/log.
- Description:
-   ("hooks failed",
-   { exns =
-       [ { exn =
-             "Unix.Unix_error(Unix.ENOENT, \"unlink\", \"/tmp/build2b91e7.dune/extunixac1694write\")"
-         ; backtrace =
-             "Raised by primitive operation at Stdune__path.clear_dir.(fun) in file \"src/stdune/path.ml\", line 1163, characters 15-34\n\
-              Called from Stdlib__list.iter in file \"list.ml\", line 110, characters 12-15\n\
-              Called from Stdune__path.clear_dir in file \"src/stdune/path.ml\" (inlined), line 1178, characters 20-45\n\
-              Called from Stdune__temp.clear_dir in file \"src/stdune/temp.ml\", line 84, characters 2-20\n\
-              Called from Stdune__exn_with_backtrace.try_with in file \"src/stdune/exn_with_backtrace.ml\", line 9, characters 8-12\n\
-              "
-         }
-       ]
-   })
- Raised at Stdune__code_error.raise in file "src/stdune/code_error.ml", line
-   9, characters 30-62
- Called from Stdlib.at_exit.(fun) in file "stdlib.ml", line 553, characters
-   62-65
- Called from Stdlib.do_at_exit in file "stdlib.ml" (inlined), line 556,
-   characters 20-39
- Called from Stdlib.exit in file "stdlib.ml", line 559, characters 2-15
- Called from Main in file "bin/main.ml", line 264, characters 11-17
- 
- I must not crash.  Uncertainty is the mind-killer. Exceptions are the
- little-death that brings total obliteration.  I will fully express my cases. 
- Execution will pass over me and through me.  And when it has gone past, I
- will unwind the stack along its path.  Where the cases are handled there will
- be nothing.  Only I will remain.
[ERROR] The compilation of extunix failed at "/home/opam/.opam/opam-init/hooks/sandbox.sh build dune build -p extunix -j 47 @install @runtest".

eDIT: this looks like a dune issue though

[camelus]

Commit: 766a299

A pull request by opam-seasoned @ygrek.

☀️ All lint checks passed 766a299

• These packages passed lint tests: extunix.0.3.1

---

☀️ Installability check (+1)

• new installable packages (1): extunix.0.3.1

[mseri]

I have notified dune maintainers and confirmed that the issues is with dune being excessively strict when deleting temp files. I think this can be merged, @kit-ty-kate do you agree?

[avsm]

I filed a dune issue, looks good to merge!

[kit-ty-kate]

This release still had some issues. Our CI had some problems and did not test the revdeps without saying anything.
None of the versions of the devkit package compile. There seems to be some issues when linking:

# ocamlfind ocamlmktop -linkpkg -thread -thread -package yojson -package yojson -package libevent -package libevent -package netstring -package netstring -package lwt.unix -package lwt.unix -package extlib -package extlib -package pcre -package pcre -package zip -package zip -package curl -package curl -package curl.lwt -package curl.lwt -package extunix -package extunix -package bytes -package bytes control.cmo exn.cmo extEnum_merge.cmo extEnum.cmo factor.cmo lwt_engines.cmo prelude.cmo files.cmo logger.cmo stre.cmo devkit_ragel.cmo time.cmo log.cmo action.cmo nix.cmo async.cmo bit_struct_list.cmo cache.cmo extArg.cmo memory.cmo signal.cmo daemon.cmo httpev_common.cmo digest_auth.cmo exn_lwt.cmo mtq.cmo extThread.cmo fastBase64.cmo gzip_stream.cmo gzip_io.cmo htmlStream.cmo var.cmo web.cmo httpev.cmo idn.cmo pid.cmo logstash.cmo lwt_flag.cmo lwt_mark.cmo lwt_util.cmo mVar.cmo network.cmo parallel.cmo persist.cmo php_serialize.cmo reader.cmo static_config.cmo unsafeBitSet.cmo -o devkit.top
# + ocamlfind ocamlmktop -linkpkg -thread -thread -package yojson -package yojson -package libevent -package libevent -package netstring -package netstring -package lwt.unix -package lwt.unix -package extlib -package extlib -package pcre -package pcre -package zip -package zip -package curl -package curl -package curl.lwt -package curl.lwt -package extunix -package extunix -package bytes -package bytes control.cmo exn.cmo extEnum_merge.cmo extEnum.cmo factor.cmo lwt_engines.cmo prelude.cmo files.cmo logger.cmo stre.cmo devkit_ragel.cmo time.cmo log.cmo action.cmo nix.cmo async.cmo bit_struct_list.cmo cache.cmo extArg.cmo memory.cmo signal.cmo daemon.cmo httpev_common.cmo digest_auth.cmo exn_lwt.cmo mtq.cmo extThread.cmo fastBase64.cmo gzip_stream.cmo gzip_io.cmo htmlStream.cmo var.cmo web.cmo httpev.cmo idn.cmo pid.cmo logstash.cmo lwt_flag.cmo lwt_mark.cmo lwt_util.cmo mVar.cmo network.cmo parallel.cmo persist.cmo php_serialize.cmo reader.cmo static_config.cmo unsafeBitSet.cmo -o devkit.top
# File "_none_", line 1:
# Error: Required module `ExtUnix' is unavailable
# Command exited with code 2.
# make: *** [Makefile:39: top] Error 10

[mseri]

Argh. I had not yet realised the CI issue back then. There may actually be a number of other packages with broken revdeps

[kit-ty-kate]

Luckily this seems to be the only one for this PR: http://check.ocamllabs.io/diff/1617127610-86c6c9a7aa07f22c4aa560fa3003a920d2a8f13f..1618491516-7af4b295dc2de0b2b594c05ab4af40ef963bbc5c
There are two old packages also broken from the latest release of h2 but that's minor.

[ygrek]

it looks like only old versions of devkit are affected, it is not worth investigating why, I will just drop them with the next release of devkit.

[kit-ty-kate]

oops sorry my bad, indeed I missed devkit > 1.x for some reason when looking at their available versions. I'll send a PR to add the appropriate constraint to those old versions right away