Re: svn commit: rev 1124 - trunk/subversion/libsvn_wc

On Thu, Jan 31, 2002 at 03:56:28PM -0600, Karl Fogel wrote:
> Greg Stein <gstein@lyra.org> writes:
> > Really?! Crap. I asked at one point and somebody blamed APR.
> >
> > Well, feh. We definitely must change it then.
> >
> > ISO 8601, baby!
>
> +1 -- I'm all for it.
>
> But I don't think this should be done by changing either of these
>
> const char *svn_time_to_nts (apr_time_t when, apr_pool_t *pool);
> apr_time_t svn_time_from_nts (const char *timestr);
>
> Right now they implement a nice read-write invariance on apr_time_t,
> and it seems foolish to just throw that away. Converting apr_time_t
> to ISO 8601 and back can lose information.

Can it? Are you *sure* about that? Or just guessing? :-)

Ahem.

Here is a valid ISO 8601 date:

    2001-11-21T09:15:56.423888Z

That is the ISO 8601 form for a date that I just extracted from a
.svn/entries file.

> Probably those functions belong in APR
>
> const char *apr_time_to_nts (apr_time_t when, apr_pool_t *pool);
> apr_time_t apr_time_from_nts (const char *timestr);

Yes. APRUTIL's "date" package, to be precise. We'd name them different,
though, to reflect their 8601 nature.

> since Subversion itself doesn't really need that r-w invariance. Then

But we'll have it.

So our internal form will be an apr_time_t. Our marshalled form is ISO8601.
The two are interchangeable.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org