Please allow smaller WUs on FAHClient

[JimF]

Now that you mention it, whatever happened to the "streaming" experiment? As I recall, it broke up work units into short pieces and continually uploaded results while downloading new work. I guess it didn't work, but I never heard the final results.

[bruce]

It worked, though I can't say how well it worked compared to the existing system.

The biggest drawback (IMHO) was that it required a total rewrite of the server code ... being incompatible with the well-established protocols, code, and methodology. ... plus nobody ever figured out how to count points in a way that was close to the points that had been accumulated up to that point so it probably would have required a reset-to-zero restart ... sort of like the Sony cell-phone client required. ANY suggested change to the points system would have been grossly unpopular with the Donors.

[JimboPalmer]

If certain projects were broken into shorter time frames:
and listed as small packets, just to reuse old 56k modem code.
and given a QRB of next to nothing, so 'real' users were not tempted by them.

That could provide a group of WUs for those convinced they need short time frames.

There would still be a huge increase in the stats database, unless the PPD is crippled to make sure no one not needing them bad, uses them.

I have no idea why any researcher would choose to do this to his/her project.

[bruce]

NaCl does just that. WUs are all short. The QRB is disabled.

I'm not sure there's any beneift of using the "old 56k modem code. That will still work AFAIK but I don't have a modem so I can't test it.

The only improvement somebody might suggest is being able to run NaCl without the overhead that Google Chrome adds to it.