tools/parse_specialcasing sometimes fails

[kit-ty-kate]

While testing several variants of ocaml for the new upcoming OCaml 4.07 I noticed a weird failure happening inconsistently when trying to build camomile.1.0.1:

#=== ERROR while compiling camomile.1.0.1 =====================================#
# context              2.0.0~rc | linux/x86_64 | ocaml-variants.4.07.0+beta2+flambda | file:///home/opam/opam-repository
# path                 ~/.opam/4.07.0+beta2+flambda/.opam-switch/build/camomile.1.0.1
# command              ~/.opam/4.07.0+beta2+flambda/bin/jbuilder build -p camomile -j 63
# exit-code            1
# env-file             ~/.opam/log/camomile-1-882bca.env
# output-file          ~/.opam/log/camomile-1-882bca.out
### output ###
# File "Camomile/jbuild", line 4, characters 0-96:
# Warning: File installConfig.ml is both generated by a rule and present in the source tree.
# As a result, the rule is currently ignored, however this will become an error in the future.
# To keep the current behavior and get rid of this warning, add a field (fallback) to the rule.
#    ocamlyacc Camomile/internal/uReStrParser.{ml,mli}
# 4 shift/reduce conflicts.
# parse_specialcasing Camomile/database/special_casing.mar (exit 2)
# (cd _build/default/Camomile && ./tools/parse_specialcasing.exe database unidata/SpecialCasing.txt)
# Fatal error: exception Not_found

Full log here: http://51.145.134.72:8080/4.07.0+beta2+flambda-Oclassic/bad/zed.1.6 I have no idea what is causing this since it only happened once and I couldn't reproduce it afterwards.

cc @mshinwell in case this has something to do with -Oclassic used here (but I cannot reproduce it, I'll try one more time)

[yoriyuki]

Could you take a backtrace for this exception?

Looking to the full log, parse_unidata is not invoked before parse_specialcasing, which shouldn't. Then, parse_specialcasing cannot find a necessary data and throws Not_found.

But I think the dependency is correctly described in jbuild. I don't know what's going on here.

[kit-ty-kate]

@yoriyuki as i said I didn't manage to reproduce it yet but as soon as I do I'll send you the backtrace.

@rgrinberg has the (deps (alias (database))) semantics been changed recently ?