quark added a comment.
I'm trying to write a version that decouples the rendering logic from the
data. But it's trickier than I thought (ex. issue5699). I guess some
refactoring needs to be done. Meanwhile I think it would be good enough if you
can avoid passing `fm` around.
INLINE COMMENTS
> rebase.py:556
> + fm = None
> + if repo.ui.configbool('experimental', 'showhashchanges'):
> + fm = repo.ui.formatter('rebase', opts)
I think it's unnecessary to have a config option - there is no behavior change
if users use rebase as before.
> rebase.py:560
> clearrebased(ui, repo, self.destmap, self.state, self.skipped,
> - collapsedas, self.keepf)
> + collapsedas, self.keepf, fm=fm)
> +
Instead of passing `fm` around, it would be cleaner if the content could be
generated, like:
# "fm" is not passed elsewhere
with ui.formatter('rebase', opts) as fm:
fm.startitem()
fm.data(nodechanges=calculatenodechanges(mapping))
(`node` seems to be more consistent with existing terms than `hash`)
But this might be difficult without the `args` templatekeywords context.
> test-rebase-base.t:418
>
> +Enabling obsolete markers
> +
The `-base.t` test was intended for testing `--base` flag. We probably want a
new `.t` file like `test-rebase-templates.t`.
> test-rebase-base.t:462
> + $ hg rebase -s 8 -d 2 -T '{hashchanges|json}' -q
> + {"38bd5f90ba6a": ["28cb79571ec7"], "583565ab89ac": ["89a178738706"]}
> (no-eol)
> +
For JSON output, it's usually for automation and better to use full hashes.
REPOSITORY
rHG Mercurial
REVISION DETAIL
https://phab.mercurial-scm.org/D934
To: pulkit, #hg-reviewers
Cc: quark, dlax, durin42, mercurial-devel
_______________________________________________
Mercurial-devel mailing list
Mercurial-devel@mercurial-scm.org
https://www.mercurial-scm.org/mailman/listinfo/mercurial-devel
