Re: RFC: date parser rewrite

On Mon, 2003-12-15 at 15:21, kfogel@collab.net wrote:
> Greg Hudson <ghudson@MIT.EDU> writes:
> > +1. Since I think we only use this kind of date-parsing in one place
> > (the -r '{DATE}' argument), I don't think this change is very
> > destabilizing, even though it churns a lot of code.
>
> Don't we use it anywhere the -r argument is accepted? Maybe it's most
> commonly used with 'svn log', but it *could* be used by any command
> that accepts -r (with or without a range).

Yes, thus "one place (the -r '{DATE}' argument)". I did not mention the
"log" command anywhere in the text you quoted.

> The current situation is certainly not ideal, but it is compatible
> with CVS dates.

I don't think that's terribly important.

> Do we really need a whole new implementation for 1.0? If the bug is
> that we don't support ISO 8601 dates

No, that's not the bug. The bug is that getdate.y supports a massive
and poorly-defined set of date formats (I posted several egregious
examples earlier), and we don't want to be trapped into supporting a
superset of those formats for all time.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Dec 15 22:24:29 2003