Dear Harvesters, Hi Daniel,
I tought I just drop in oe or two thoughts on this subject:

Firstly I thought _becoming_ a Harvester don't seem so hard to do, it's
the work itself that is the hard part (and thanks for that). So if you
Celeste folk are really that interested in your stuff (and it seems to
evolve very neatly) then why can't you just assign the job of a special
'Celeste-Harvester' to someone?! I think this is much more feasible in
the near future than a dedicated repository (i may be wrong).

Secondly _please_please_please_ spare us (the list) with all of the
comments why this_and_that code snippet was not included in the update
stream. I think the solution with the wiki-page is better than nothing
and _much_ better than the additional 'noise' on the list.

Just my 2cents

Doug Way wrote:
> 
> danielv at netvision.net.il wrote:
> 
> > Dear Harvesters,
> > First of all, I also want to thank you for doing this work, it's really
> > important.
> >
> > And now a request - is it possible to make sure the you give some
> > indication as to the fate and faults of anything considered for
> > inclusion but rejected?
> >
> > I'm assuming that you scan everything submitted. Some things have
> > problems, some things seem to be of no interest to the public, and so
> > forth. As a contributor, I'd like to know which felled my contributions,
> > and seeing as you already review everything, it should not be hard to
> > get this feedback to the authors. I know this is more work, and you're
> > already undertaking quite a bit, but I think it's really important.
> 
> We have discussed this, and I agree that this feedback is important.
> 
> For now, the way to find out the status of past submissions is to check the sqfixes tables site that Bob mentioned. (http://209.143.91.36/super/415)  I was going to mention this earlier, but Bob beat me to it. :)  This site contains tables with all of the submissions in the left column, and the current status in the right column.
> 
> This might not be quite as convenient (for the authors) as getting feedback mailed directly to each author, but I think it's a significant step forward... much better than no feedback (the previous situation).
> 
> On the other hand, even though most submissions have some sort of status or comment filled in, as Bob mentioned there are still quite a few blank ones with no feedback.  Much of that was probably due to there being a lot (~500) of old submissions to wade through this time... several months' worth.  I think things will get better as we get caught up, and also as we have time to improve the process.
> 
> > Otherwise people have to either ask (which is more of a bother) or
> > assume things that are probably not encouraging them to contribute
> > further...
> >
> > A mail to the list enumerating the issues with each submission not
> > accepted seems like a good way to do it.
> >
> > This could also be a spur for people to talk about what submissions they
> > do and don't care about, and for fixing/improving those that merit it.
> 
> That's not a bad thought.  I'm not sure we're at the stage of being able to do that just yet, though...
> 
> > I think this feedback/disscussion/improvement is a really important part
> > of an open source community, and making decisions about contributions
> > more visible would help it a lot.
> >
> > On another side of the same coin -
> > There are a few of us (Lately - Mike R., Lex, Leandro and myself,
> > sometimes other members too) that are maintaining and enhancing Celeste
> > over time. On one hand we'd like to tighten the release loop in order to
> > have a clear platform to work on, and on the other hand we know you
> > Harvesters are busy, so we prefer to give you less work, not more.
> >
> > We were thinking of starting to maintain Celeste on a repository, so we
> > can integrate changes into one official version independently of the
> > Harvesting process. This would give us and Celeste users a faster
> > evolving package, and we hope you will find it easier to simply take our
> > integrated changes from time to time, and update the image accordingly.
> >
> > Does this sound reasonable, am I missing something, or do you have other
> > ideas?
> 
> This sounds like a good idea to me.  (I purposely skipped looking at any Celeste-related fixes myself, partly because I don't know much about Celeste, but also because I knew that there were probably others working on maintaining it...)  We can work on coordinating this.
> 
> - Doug Way
>   dway at riskmetrics.com