Re: Mark/seek/reset problems in svn_subst_stream_translated()

On Thu, Apr 15, 2010 at 05:36:36PM +0100, Julian Foad wrote:
> I (Julian Foad) wrote:
> > The "line filter" and "line transformer": I don't think these fit the
> > stream abstraction at all well. Streams don't have "lines" as a primary
> > concept. Without going into the details, I think it can't play nicely
> > with plain reads and writes, and I doubt it's properly integrated with
> > mark/seek.
>
> (That could be construed as a very rude thing to say, but I just meant
> that's something I think will be quite difficult to get right and it's
> still a doubt in my mind because I haven't studied it yet.)

I agree that the current way line filtering is done is suboptimal.

Maybe we should move all of that into functions wrapping
svn_stream_readline() within the diff parsing and patch code?

Stefan

-- 
Sent from my Nokia 1100