SieVe Automation

[Death]

well.

how 'bout subj?

something like clients connect to server, obtain some range and submit results back to server.

here http://wiki.seventeenorbust.com was some wishes that client v.3 can do sieveing, PRPing and 17oB! for your choice.

but as for current version - how long it takes to develop something like this?

and is it worth to do this?

[Mystwalker]

Well, I basically already developed this, but it uses SoBSieve and is not fully tested. I hope I find the time to finish that work...

[Death]

Originally posted by Mystwalker
Well, I basically already developed this, but it uses SoBSieve and is not fully tested. I hope I find the time to finish that work...

well how about server part?

[Mystwalker]

Done - though a rewrite wouldn't hurt. If someone likes, I can provide the source code. It's written in Java. The server needs a MySQL database.

[Death]

Originally posted by Mystwalker
Done - though a rewrite wouldn't hurt. If someone likes, I can provide the source code. It's written in Java. The server needs a MySQL database.

need to hear from admins.
server can submit small ranges like 10M and wait for factors or for "no factors found" message. this can keep overall range completed more smoothly.

[Nuri]

I guess you meant 10G, right?

Roughly speaking, 10M takes 20 to 40 seconds. Server would go .


BTW, most of my machines work off-line, and I personally would prefer if manual reservation system is also preserved.

[Death]

Originally posted by Nuri
I guess you meant 10G, right?

Roughly speaking, 10M takes 20 to 40 seconds. Server would go .


BTW, most of my machines work off-line, and I personally would prefer if manual reservation system is also preserved.

I mean 10 units that you use for start and end sieve point... =))

ofcourse, and server must know this ranges.

[Mystwalker]

I designed server & protocol such that arbitrary ranges can be reserved.
I just need to write a little enhancement for declaring part of a range as completed. Hopefully, I can do some work on Sunday...

[MikeH]

BTW, most of my machines work off-line, and I personally would prefer if manual reservation system is also preserved.

Me too.

[Mystwalker]

Originally posted by Nuri
BTW, most of my machines work off-line, and I personally would prefer if manual reservation system is also preserved.

I don't see a problem here.
One could simply reserve a quite large range (e.g. 10T) for automately distributed sieving.
Well, I already kinda did this last summer - maybe someone remembers "secretTesting"?

[Nuri]

Yep, how can one forget those mysterious posts.

Mystwalker, I was referring to the development of integrated client (PRP & sieve together i.e. v.3). If we go to full automated server reservation system, what I'm afraid is that the server would decide the length of range to issue to the client and the user would not be able to reserve, say 10T.

One other thing I tried to emphasize was that we should still be able to manually distribute the ranges we reserved to our off-line boxes. It's OK as long as we can at least do that.

On the other hand, if I understand the way your program works correctly, I also have no problem with the way it works.

[cjohnsto]

Originally posted by Nuri
One other thing I tried to emphasize was that we should still be able to manually distribute the ranges we reserved to our off-line boxes. It's OK as long as we can at least do that.

On the other hand, if I understand the way your program works correctly, I also have no problem with the way it works.

Well it shouldn't be too hard to write a stub client that downloads a range and just prints it out or creates a file to give to the offline client, and conversely takes a results file and passes it back to the server.

[jjjjL]

the main goal in integrating the sieve into the client is not so much to deepen the sieve as it is to make the client work on a broader range of systems since it may be easier to port and rebuild as a plugin.

we don't intend to replace the main sieve effort, only supplement it.

Cheers,
Louie

[Nuri]

Excellent.

[Mystwalker]

Ok, I finally managed to get the client wrapper working with proth_sieve - with some annoying occurrence on the way (like starting several hundred proth_sieve instance within a few seconds - noteworthy: every instance had to connect to the server before the next one starts ).

Now I'd like to give the system a broader test (and tidy up some mess etc.) - which puts me in the need of a system hooked up to the internet (preferably) 24/7 and reachable with a "static" hostname (which means either static IP, static hostname or dynDNS service).
It has to be able to run Java Applications.
Access to a (esp. local) MySQL database would be a big plus.

Our own home server in the basement seems to close the port of the server after some time (at least I hope it's the server and not Java itself...).

Ok, long story short:
If there is someone with the appropriate resources and willing to help, just answer in this thread or PM me. Thanks!


btw.:
It still takes some time for a release to the wild - but a (more or less) closed beta test should be manageable within a week or two.