Re: [PATCH v3 2/3] fast-import: add option command

["Shawn O. Pearce" <spearce@spearce.org>]

Sverre Rabbelier <srabbelier@gmail.com> wrote:
> On Fri, Aug 14, 2009 at 08:37, Shawn O. Pearce<spearce@spearce.org> wrote:
> > Sverre Rabbelier <srabbelier@gmail.com> wrote:
> >> +`option`
> >> +~~~~~~~~
> >> +Processes the specified option so that git fast-import behaves in a
> >> +way that suits the frontend's needs.
> >> +Note that options specified by the frontend are overridden by any
> >> +options the user may specify to git fast-import itself.
> >
> > Wha? ?This disagrees with the code.
> 
> It does? It's saying that the user my override what the frontend
> specifies, which is what the current version does.

Sorry, lack of caffeine.  Rereading the docs makes sense.
 
> >> + ? ? if (!seen_non_option_command)
> >> + ? ? ? ? ? ? parse_argv();
> >
> > This is too late.
> 
> No it's not. Earlier in the patch, at the other
> 'seen_non_option_command', we call parse_argv() as well (which happens
> when a non-option command is issued). This statement is here to deal
> with options that affect an empty stream, such as 'git format-patch
> --import-marks=marks.old --export-marks=marks.new < /dev/null &&
> test_cmp marks.old marks.new'.

Oy, I missed that call.  I looked for it but gave up too soon I guess.
 
> > So yea, I really do think its a good idea for command line options
> > to override stream options, despite what Dscho may think. ?:-)
> 
> Which is what the current version does, yes?

Yes.  :-)

-- 
Shawn.