Closed art-w closed 2 years ago
Thanks for the PR!
Wrt lifting opam options to marracheck options: our workflow until now was to use environment variables for that. Most (every?) --foo
option you can pass to opam
has an equivalent OPAMFOO
environment variable you can use instead. Defining these environment variables alsos work with marracheck (it got broken at some point in the past but should now work again).
So launching marracheck with OPAMDEBUG=1 OPAMVERBOSE=1
should work for instance.
Granted, this is sort of a hidden feature at the moment. I'm not sure how much we should try to lift (a subset of) opam options into marracheck options, or instead just mention in the README a list of opam-related environment variables that one can use and that are generally useful.
Oh nice, I completely forgot about environment variables but they solve all my issues! I've removed the lifting of opam options in favor of more explanations in the README (but feel free to close this PR, I've no attachment to what remains)
thanks! the rest of the PR is still very good to have! merging.
The main motivation was to enable opam's
--debug --verbose
to get more feedback when opam is hanging for minutes, but there are a few more settings that may be interesting:--fake
to spare a few cycles/disk space when debugging elements cover (but note that--dry-run
doesn't work as it doesn't register the switch!)--keep-build-dir
,--with-doc
,--with-test
for a slightly different usage--no-depexts
,--yes
,--no
, etc for running without human interventionOpam's
--switch
could cause conflicts with marracheck's positional argument, so the proposed fix is to use the opam argument (with an arbitrary default of4.13.1
if missing):While the extra settings are convenient, I'm not 100% happy that the
--help
becomes noisy with all the random opam options... And it's very likely that some of these options will interact badly with marracheck, although I'm really not sure which one should be forbidden / provide no value.