Re: [Quilt-dev] Time for a new release?

[Andreas Gruenbacher]

On Saturday 09 July 2005 03:34, Peter Williams wrote:
> Andreas Gruenbacher wrote:
> >  - I would rather make this the header command. We are usually calling
> > what's above the actual patch the header, no?
>
> Yes, but this command only extracts/sets the description component of
> the header which may also contain diffstat data.  I.e. it's all about
> the MANUALLY created components of the header file.

I don't agree with you that the command should special case the diffstat part.

> I agree that 
> "description" is rather a long name and that a shorter one would be
> preferable, e.g. "descr"?.

As Jean mentioned, the length of the name is not an issue.

> The "files" command goes part way towards providing the diffstat data
> perhaps it could be modified with an option to add diffstat data to its
> output?

I don't think it should.

> >  - I don't see why it's useful or necessary to remove the diffstat part.
> > Any reasons? Presumably...
>
> That's automatically generated and the description isn't.  Omitting it
> means that the user doesn't have to worry about reproducing it when the
> change the description using the -s option.

When the user removes the diffstat output, the next refresh will reinroduce it 
at the end of the header. It could also be somewhere in the middle of the 
header too though, and then this new command would start to behave strangely.

I would be fine with a compromise: make the diffstat-stripping optional.

> This would become even more important if we were to add an append (e.g.
> -a) option to append text to the existing description as an optional
> alternative to the -s option.  Arguably, this would be more useful to
> somebody using quilt from the command line than the -s option would be.
>   Appending comments after the diffstat data would probably break
> "refresh".

It wouldn't break refresh, no.

> >  - Is removing whitespace in the header really needed?
>
> Neatness counts :-)
>
> > I don't think anybody
> > cares much about whitespace in the header at all; it's annyoing in the
> > actual patch though.
>
> To me whitespace at the end of lines is annoying everywhere :-)
>
> Unfortunately, some editors are difficult to configure so that they
> don't leave them dotted around and this provides an opportunity to get
> rid of them.

Doh. Everything and the kitchen sink.

> >  - The command must not touch $patch~refresh.
>
> I hope I didn't do that.  If I did it was unintentional.  Does this mean
> that the access times on the patch file have to be remain unaltered?  If
> so how is that accomplished?

It's fine to touch the patch. $patch~refresh indicates that the patch needs 
refreshing before it can be popped, which clearly is not the case here.

-- Andreas.