Re: [PATCH v2 0/2] remote-bzr: couple of fixes

From: Felipe Contreras <felipe.contreras@gmail.com>
Date: May 5, 2013, 18:42 UTC
Message-ID: <CAMP44s1D7LOhDGkZguosPiXyuJ5cP2hmgq4AWagwadrJYK1Pgg@mail.gmail.com>
In-Reply-To: <7vd2t5uvi2.fsf@alter.siamese.dyndns.org>

On Sun, May 5, 2013 at 1:33 PM, Junio C Hamano <gitster@pobox.com> wrote:

Show 10 quoted lines

> Felipe Contreras <felipe.contreras@gmail.com> writes:
>
>> The previous version had an indentation bug (did I mention I hate python?).
>>
>> A few fixes to be applied on top of the massive changes already queued. Nothing
>> major.
>
> [2/2] may not matter much in the context of my tree (people would
> use post 1.8.2 fast-export if they are using remote-bzr from 1.8.3
> from my tree ;-),

Maybe, but if even if they have the latest git, pushing a tag will fail miserably, and with the patch it would fail nicely :)

Show 7 quoted lines

> but [1/2] sounds like it is a good thing to have
> in 1.8.3 (not "on top of that 'massive' series").
>
> Assuming the "otherwise some version of bzr might barf" problem is
> that repo.generate_revision_history() in those versions may not
> apply str() to its first parameter and the caller is expected to
> pass a string there, or something?

No, there's no change to repo.generate_revision_history(), because we already convert the elements of the array to strings, it's the other callers of Marks::to_rev() that see a change, namely code that pushes to a remote, I think.

And BTW, they are already strings, but unicode strings, because they come from a json file, somehow bazaar doesn't like that, but it works fine in my machine without the patch. Shrugs.

Also, the emacs developers seem to be fine with all these changes, there's only one patch pending that I need to cleanup.

Cheers.

-- 
Felipe Contreras