Re: vcs_info on a hg repository with a git subrepository; actionformats works ok, formats does not

Re: vcs_info on a hg repository with a git subrepository; actionformats works ok, formats does not

[Frank Terbeck]

[moving thread to -workers]

Seth House wrote:
> On Mon, Sep 17, 2012 at 9:57 AM, Seth House <seth@eseth.com> wrote:
[...]
>> The --debug flag is used to get hg to output the full sha.
>
> Relying on --debug here is not ideal. Something like this was bound to
> happen eventually. Unfortunately, I don't believe there's a way to
> retrieve all info of interest to VCS_Info with a single call to
> Mercurial. (Invoking the Python interpreter is too slow to do more
> than once while drawing the shell prompt.)

Indeed. :-/

> I have done quite a bit of experimenting, searching, and asking on
> this.
>
> I propose one of the following:
>
> 1) Alter the hg backend to work without the full 40-char hash.
>
> The ``hg id`` command will always return a unique hash. The only thing
> the full hash is being used for in the hg backend is comparisons with
> MQ patches and that can be changed to be a left-match instead. We can
> document why the full hash isn't available and if users are expecting
> 40-chars we can give an example of how to pad the value with spaces.
> This changes existing behavior.
>
> 2) Continue using --debug and add a workaround for this issue.
>
> It's plausible that the output of interest will always appear at the
> bottom of the debug output. Mercurial does have a
> backward-compatibility policy for its output formats but this does not
> extend to debug output (obviously).
>
> FWIW, my personal preference is the first option as I feel the 40-char
> hash is rarely desired or useful in a shell prompt.
>
> Thoughts?

Hm. Ideally, I'd like there to be the least possible impact on existing
setups. But then and again having `--debug' in there was always
bothering me a little.  I'm undecided.

I know, however, that I don't want to introduce another style again to
make this optional.

I'll think about it until tomorrow. If a user of the hg backend has a
strong opinion on this, speak up.

Regards, Frank

-- 
In protocol design, perfection has been reached not when there is
nothing left to add, but when there is nothing left to take away.
                                                  -- RFC 1925

Re: vcs_info on a hg repository with a git subrepository; actionformats works ok, formats does not

[Frank Terbeck]

Frank Terbeck wrote:
[...]
>> 1) Alter the hg backend to work without the full 40-char hash.
>>
>> The ``hg id`` command will always return a unique hash. The only thing
>> the full hash is being used for in the hg backend is comparisons with
>> MQ patches and that can be changed to be a left-match instead. We can
>> document why the full hash isn't available and if users are expecting
>> 40-chars we can give an example of how to pad the value with spaces.
>> This changes existing behavior.
[...]
>> FWIW, my personal preference is the first option as I feel the 40-char
>> hash is rarely desired or useful in a shell prompt.

After some thought, I agree. Relying on debugging features to remain
stable is not a good idea. Seth: Are you willing to send a patch?

Regards, Frank

Re: vcs_info on a hg repository with a git subrepository; actionformats works ok, formats does not

[Seth House]

On Wed, Sep 19, 2012 at 12:34 AM, Frank Terbeck <ft@bewatermyfriend.org> wrote:
> After some thought, I agree. Relying on debugging features to remain
> stable is not a good idea. Seth: Are you willing to send a patch?

Can do.