Loosing structures when changing proteins

[pointwood]

Hi,

I know this has been asked before, but I haven't seen an answer to the question yet.

Why is it that it is not possible to upload generated structures after you have changed to a new protein?

Some people have lost quite a lot of work (me included) because you changed to a new protein on a sunday or something, without giving them any time to upload their work.

If at all possible, it would be nice if it could be made possible to upload work from the previous protein, even if it where just for a limited time. No matter what, I think you should try to find a better way to manage this and then clearly describe how protein changes happens and what the consequences are.

[Paratima]

I agree. Consider: You have n boxen crunching like mad. When you cross the magic boundary, you have n-1 boxen still crunching! Some of these will have just uploaded and resumed work. They are going to waste an entire work cycle, which for the slower machines can be several hours. :(

At the very least, all results must be successfully uploaded and removed from user disk drives. At the best, users would get credit for those WUs, say, those returned within 24 hours, whether you actually capture/use them or not. If you use exactly 1 billion WUs, but give credit for another hundred thousand, what's the problem? ;)

[FoBoT]

howard/brian the fist did answer this in another thread

http://free-dc.org/forum/showthread.php3?s=&threadid=526


any work left un-sent when we switch proteins is effectively lost. It would be nice if we could get that extra little bit but this is simply not possible with the manner in which our hardware and server software have been implemented (without going into boring technical details..)
If you have machines not on the network, just try to keep an eye when we get close to a full thermometer and do an 'upload dump'.

so it is important to watch when we approach the completetion of a protein

if it isn't possible to be at your PC's during the switch, another stategy would be to upload your work and then switch to another project temporarily until after the new protein is in place and then put your boxen back onto DF

[Paratima]

Sorry, missed (or read & forgot) the other thread.

I still think it's a worthwhile target for a future mod, and...

I amend my antipenultimate sentence to read:

"At the very least, all prior results should be removed from the user's disk without requiring manual intervention." :cool:

[Mobocastor]

I think Distributed Folding should take those old results whether we have changed proteins or not. I had a ton of folds ready to be sent at the time of the change. I couldn't use the auto update because of this so I had to delete the entire directory to get those files off my disk after downloading the new protein from the website manually and reinstalling. I don't want worthless files accumulating on my drive after every protein change.

[1fast6]

I don't want worthless files accumulating on my drive after every protein change.

and I don't want wasted work... which is what this is...

I want my donated cycles to count... regardless if I'm around... or paying attention... or whatever...