Re: [PATCH 5/7] remote-helpers: add support for an export command

From: Sverre Rabbelier <srabbelier@gmail.com>
Date: Mar 29, 2010, 17:28 UTC
Message-ID: <fabb9a1e1003291028x45f0c570v4c10eadf37fc6a77@mail.gmail.com>
In-Reply-To: <alpine.LNX.2.00.1003291256020.14365@iabervon.org>

Heya,

On Mon, Mar 29, 2010 at 11:21, Daniel Barkalow <barkalow@iabervon.org> wrote:

> It would be good to have documentation on what the helper should do doing
> here; it's relatively difficult to tell from this code.

Yes, I plan to send a documentation patch as well but wanted to get this out the door first (since Documentation patches don't have to cook in next).

> I think that it's getting some options (import and export marks) back from
> the helper in a kind of ad-hoc and unclear fashion, and it should be doing
> something more appropriate.

It's not really that ad-hoc or unclear. When the helper receives the export command it should write two lines, the first being the location where it should write the marks to, the second being the location where it should import marks from. Either line may be empty, meaning marks should not be written or read respectively.

> (I'm not sure what the more appropriate thing
> is, or really what the requirements are here as far as getting information
> around, but I think it needs to be discussed.)

The import command solves this by printing 'feature import-marks' and 'feature export-marks' lines on the stream, but that doesn't work here since fast-export doesn't read from stdin. I don't think this needs a separate command.

-- 
Cheers,

Sverre Rabbelier