[regression] annotate --follow no longer lists file name (was: Find a regression, win a t-shirt!)

Steve Borho steve at borho.org
Mon Feb 15 20:54:02 UTC 2010


On Mon, Feb 15, 2010 at 2:34 PM, Thomas Arendsen Hein
<thomas at intevation.de> wrote:
> * Dirkjan Ochtman <dirkjan at ochtman.nl> [20100215 13:02]:
>> On Mon, Feb 15, 2010 at 12:39, Thomas Arendsen Hein
>> <thomas at intevation.de> wrote:
>> > I think --follow automatically should imply --file.
>>
>> I disagree, they should be orthogonal. We changed -f to make it easier
>> for commandline users to adopt,
>
> If it is only for the command line users, --follow does no longer
> need to be kept. But if some automatic tool uses e.g.
> "hg annotate --follow --number" and expects the output to include
> the filename, it will simply break until someone adds --file or
> changes the parsing code.

Exhibit A:

http://bitbucket.org/tortoisehg/stable/changeset/4d0051b2f56b/
http://bitbucket.org/tortoisehg/stable/changeset/8fd3cc10d275/

> Yes, this is only a small change, but should not be done without
> some time to adjust the tools. (at least one major release?)
>
>> but the --file behavior still doesn't
>> seem very useful to me (if it doesn't have -n as well).
>
> Sure, and --follow should die some day, but not from 1.4.3. to 1.5.0.
>
>> > Additionally --follow should print DEPRECATED, or not?
>>
>> That might be a good idea, though I don't think that's a regression.
>
> I'm still not convinced, and I would have reported this incompatible
> change without being motivated by the t-shirt competition, too :)

It's not so bad in the grand scheme of things, I just don't like
discovering these kinds of changes when functionality breaks.  I've
subscribed to the API Changes wiki page, but this one didn't appear
there.  Is there another wiki page that tracks command line argument
changes on the default branch?  If not, why not?

--
Steve Borho



More information about the Mercurial-devel mailing list