[PATCH RFC] rename manifest command to files

[Dan Villiom Podlaski Christiansen]

On 30 Apr 2011, at 10:17, Adrian Buehlmann wrote:

> On 2011-04-30 09:58, Dan Villiom Podlaski Christiansen wrote:
>>
>> On 30 Apr 2011, at 09:04, Adrian Buehlmann wrote:
>>
>>> On 2011-04-30 08:42, Adrian Buehlmann wrote:
>>>> On 2011-04-29 22:42, Adrian Buehlmann wrote:
>>>>> # HG changeset patch
>>>>> # User Adrian Buehlmann <adrian at cadifra.com>
>>>>> # Date 1304106171 -7200
>>>>> # Node ID 2a66906672c124567829abdc67e4f81b8d635cef
>>>>> # Parent  98f79a5c308665e1ce1b020bb66ed37f47f102e6
>>>>> rename manifest command to files
>>>>
>>>> It might be worth noting that darcs has a 'darcs show files'  
>>>> command
>>>> with (!) an alias 'darcs show manifest' (they call it an alias
>>>> themselves [1], even though it seems to have slight differences in
>>>> behavior compared to their 'show files').
>>>
>>> Heh. Even better :-). Their changelog [2] says for the darcs  
>>> 2.0.0pre2
>>> release:
>>>
>>> <quote>
>>> The subcommand 'query manifest' is renamed to 'show files', ...
>>> </quote>
>>>
>>> [2] http://darcs.net/releases/branch-2.5/NEWS
>>
>> Rather than adding a new command, wouldn't it be worthwhile to  
>> combine
>> locate and manifest into one command?
>
> I'd rather like to keep this discussion focused on what I proposed:
> renaming the *current* "manifest" command to "files", that is, having
> "files" in the list of commands that are shown on "hg help" instead of
> "manifest" and keeping the command name "manifest" as an alias and for
> backward compatibility.
>
> In this patch (and thread), I have *not* proposed to add a new  
> command.

Well, we still have two commands, currently; locate & manifest. They  
do *mostly* the same things. Namely:

$ hg manifest | tail -n1
tests/tinyproxy.py
$ hg manifest -v | tail -n1
755 * tests/tinyproxy.py
$ hg manifest --debug | tail -n1
301fa977ccd17f37133e47c1cecc4d1318ecc74a 755 * tests/tinyproxy.py

The only option 'hg manifest' takes is '-r'

The locate command works on the working copy by default:

$ hg locate | tail -n1
tests/tinyproxy.py
$ hg locate -v | tail -n1
tests/tinyproxy.py
$ hg locate --debug | tail -n1
tests/tinyproxy.py
$ hg locate --debug | tail -n1
tests/tinyproxy.py
$ hg loc -X '**.py' | tail -n1
tests/test-wireproto.t
$ hg loc -I '**.t' | tail -n1
tests/test-wireproto.t

But:

$ hg locate '**.py' | tail -n1
tinyproxy.py
$ hg loc '**.t' | tail -n1
test-wireproto.t

(That they don't all print out relative paths is probably a bug, but I  
have no idea whether we can fix it under our backwards compatibility  
guarantees.)

'locate' takes loads of options, including -r/--rev, -0/--print0, -f/-- 
fullpath, -I/--include and -X/--exclude.

So, which of these should be ‘files’? Which of them makes sense? One  
possible option is to deprecate both manifest & locate, and replace  
them with a ‘files’ command that includes the functionality of both.

(FWIW all of the commands were executed in ‘~/src/opensrc/vcs/ 
mercurial/tests’.)

--

Dan Villiom Podlaski Christiansen
danchr at gmail.com