Hi Junio,

On 10/06/2019 18:57, Junio C Hamano wrote:

Philip Oakley [off-list ref] writes:

quoted

The availability of these pattern selections is not obvious from
the man pages, as per mail thread [ref].

Provide examples.

Re-order the `git branch` synopsis to emphasise the `--list <pattern>`
pairing. Also expand and reposition the `all/remotes` options.

Split the over-long description into three parts so that the <pattern>
description can be seen.

Clarify that the `all/remotes` options require the --list if patterns
are to be used.

Add examples of listing remote tracking branches that match a pattern,
including `git for-each-ref` which has more options.

Improve the -a/-r warning message. The message confused this author
as the combined -a and -r options had not been given, though a pattern
had. Specifically guide the user that maybe they needed the --list
option to enable a remote branch pattern selection.

Signed-off-by: Philip Oakley <redacted>
---

in response to
[ref]
thread: https://public-inbox.org/git/?q=%3CCACsJy8CwY8gzeWa9kNRX3ecez1JGiQiaOknbAoU7S%2BhiXBoUGQ%40mail.gmail.com%3E

to: Git Mailing List <redacted>
cc: Duy Nguyen <redacted>
cc: Ævar Arnfjörð Bjarmason <redacted>
cc: Junio C Hamano <redacted>

This looks reasonable to me---is it ready to go even with its RFC prefix?

Yes; the RFC was only in regard of the die() message change as it 
touches real code;-)


Following the post, Stolee's coverage report, it was noted that the 
die() wasn't actually tested, so I sent a quick follow-up of a potential 
test 
https://public-inbox.org/git/f28dd5b1-fda8-cf51-5582-067a7d2c2472@iee.org/ 
which is a proper RFC...


I did not attempt to see if any other dies()'s were untested, just 
copied one that was..

Philip

(currently travelling)