On 4/24/06, Stefan Küng <tortoisesvn@gmail.com> wrote:
> David Anderson wrote:
> > Hi,
> >
> > * Stefan Küng <tortoisesvn@gmail.com> [2006-04-21 20:18:29]:
> >> Since there's some discussion about what issues could be used for SoC,
> >> I'd like to add one more:
> >>
> >> Issue #901 : http://subversion.tigris.org/issues/show_bug.cgi?id=901
> >>
> >> I think this issue is big enough to keep a student busy during the
> >> summer, but still not so big or complicated that it couldn't be
> >> finished. And it can be done in stages, which is nice because even if it
> >> doesn't get finished the parts already done can be used.
> >
> > Thanks for suggesting this. I must say I have some reservations,
> > simply because it seems like an extremely open-ended task: "Provide
> > finer grained notification". Could the issue be specified more
> > precisely, so that the task falls clearly in the SoC scope?
>
> I kind of like that it's open-ended. That way even if the student
> doesn't get to do it all, we still could use whatever (s)he finished.
>
> To make it more precise, I suggest:
>
> * define the required API (client callback, enum for progress type like
> 'files', 'bytes', 'chunks', ...)
> * locate the modules where this has to be implemented (ra modules,
> libsvn_wc, ...)
> * start with one command like checkout or import (not diffs, only
> complete files). Then expand (if there's still time in the SoC
> timeframe) to other commands like commit, update, diff.
And as soon all this has been done, the need to send a patch for
serf/neon becomes eminent. If and when that hasn't been addressed by
then, I'll try to make one.
But, justin, saying there's a reason it hasn't been done (implying
this work will be useless?) somehow gives me the idea were running
around in circles: why would serf/neon implement it if there are no
consumers anyway?
bye,
Erik.
Received on Mon Apr 24 20:09:26 2006