packaging: move most of contrib/docker to contrib/packaging/docker
contrib/ is kind of a dumping ground for all kinds of random things.
Many of the files in contrib/ are related to packaging. But sorting
out how everything is related is a lot of effort since it is all
mixed together.
This commit starts the process of refactoring everything related
to packaging so it all lives under one roof.
Most content from contrib/docker is simply moved to
contrib/packaging/docker and all references are updated.
Differential Revision: https://phab.mercurial-scm.org/D3541
setup: reformat options argument
This makes things a bit more easier to manage IMO.
Differential Revision: https://phab.mercurial-scm.org/D3540
contrib: remove fixpax.py
This was used as part of bdist_mpkg. And we stopped using bdist_mpkg
after
db5084d27df5.
Differential Revision: https://phab.mercurial-scm.org/D3539
push: continue without locking on lock failure other than EEXIST (
issue5882)
This code was added by
3f5e75c22585 "push: make locking of source optional
(
issue3684)", but EACCES isn't the only error that could be triggered by
filesystem permission. I think catching LockUnavailable is more appropriate
than testing errno value by caller.
revlog: handle errors from index_node() in nt_insert() and index_slice_del()
Same idea as in
a9d9802d577e (revlog: don't say "not found" on
internal error, 2018-05-04).
Differential Revision: https://phab.mercurial-scm.org/D3558
help: add new deprecated topic
Add a deprecated topic which lists most of the commands and options that are not
recommended anymore.
phabricator: split auth.url into the standard auth.schemes and auth.prefix
It seems better to reuse the existing function to find the proper [auth] block,
even if not all of the possible settings may be of interest.
The other callers of readauthforuri() make a trip through the password database
to fetch the user from the URI. But in the little experimenting I did here, the
username always came back as None. Since readauthforuri() wants it to make sure
that user@prefix matches user@url, it seems that parsing the URL and pulling out
the user component should be equivalent.
phabricator: migrate [phabricator.auth] to [auth]
This resurrects the warning mechanism removed in
20a4543e9a2b. I'm not worried
about the copy/paste of the code for a brief transitional period.
dispatch: unify handling of None returned by a command function
A command function may return None in place of 0 just for convenience, but
dispatch() doesn't need to inherit that property. This patch makes it be
friendly to callers.
dispatch: minor code refactor
This hopefully makes things slightly easier to read.
Differential Revision: https://phab.mercurial-scm.org/D3537