[PATCH] commitextras: move fb's extension which add extras to a commit to core

[Yuya Nishihara]

On Thu, 6 Jul 2017 14:24:08 -0700, Martin von Zweigbergk wrote:
> On Thu, Jul 6, 2017 at 10:58 AM, Sean Farley <sean at farley.io> wrote:
> > Yuya Nishihara <yuya at tcha.org> writes:
> >> On Wed, 05 Jul 2017 23:14:13 +0530, Pulkit Goyal wrote:
> >>> # HG changeset patch
> >>> # User Pulkit Goyal <7895pulkit at gmail.com>
> >>> # Date 1499274297 -19800
> >>> #      Wed Jul 05 22:34:57 2017 +0530
> >>> # Node ID 64c39cf4b61475a133f88fa6dd247ca4f3d61436
> >>> # Parent  8e6f4939a69ae8949e134d97de6b766799bca8de
> >>> # EXP-Topic fbext
> >>> commitextras: move fb's extension which add extras to a commit to core
> >>>
> >>> Facebook has this as extension which adds a flag to `hg commit` to add extra
> >>> values to a commit from command line. This patch moves that extension to core by
> >>> adding that flag and marking that as ADVANCED. Also this patch adds a function
> >>> to check if any key which is used internally can't be used.
> >>
> >> FYI, the last attempt was rejected.
> >>
> >> https://www.mercurial-scm.org/pipermail/mercurial-devel/2014-March/056889.html
> >
> > Ha, I was going to dig that message up as well.
> >
> >> I don't know if ADVANCED flag can make the "high barrier" enough to discourage
> >> an abuse of extras.
> >
> > I agree with this and with Matt's original criticism. Can this not be an
> > extension instead?
> 
> How about make a debugcommit command that adds this flag and delegates
> everything else to the regular commit command?

debugcommit seems fine.

FWIW, --extra value will need some encoding handling depending on its
type, e.g. convert labels to utf-8, but keep file path as bytes.