External Help Wanted with Development?

[Mr.Nosmo]

I was sitting reading a article about Linux when it hit me: Why don't Stanford make the code available to a few thousand programmers (maybe they have to sign a NDA) so the development-time can be improved? I'm not a programmer, but it looks like there are a few good ones around this forum...... Just idea....

[MtM]

Stanford PG has never released source because they want to keep the validation process a secret A 'few thousand' is I think a big security risk, as true, it's a science project but there are people who are only interested in points and when there is a demand, someone might sway and be 'bought'?

Other then that argument I'm all for it

[Mr.Nosmo]

True, there is a security risk, but maybe there are things that "outsiders" can help with without Stanford PG have to loose sleep over it.... The reason for me to say "a few thousand" is to have programmers to submit code that Stanford then look at and the best code can be included in the Clients or Cores...... If I could program a bit more than make my Commodore64 print "I'm the king" all over the screen, then I would be proud to have written code that helps this project......

[Ivoshiee]

It is already happening - if you look a little closer then at least one "outsider" and this forum resident is already contributing to the FAH client development.

[Mr.Nosmo]

Point taken "Ivoshiee", but how many "Outsiders" are there? And how can you "apply"?

[MtM]

Yeah I'm actually also trying something simular hehe

I'm not a programmer by profession though, and I'm limited to .net but it's a powerfull language/backend anyway.

I do think they would benefit from such a usergroup, since I know I'm pretty limited and there are many people out there with greater abilities then me and I think with atleast the motivation and enthousiasm as well.

I think btw they already have this going on, but maybe outside of this forum? I do not know the development group of for example the a2 core but I wouldn't be surprised if there where multiple groups of people working on that.

I'm trying to make an installer which can install the smp clients and gpu clients on machines with as litle user actions needed to set it up and start running. Don't get allot of support atm, not from PG and not other parties ( like Nvidia which I asked acces to the full nvapi months ago so I could enable/disable sli for folding respectively gaming if a donor wanted to stop their clients and play a game ).

I been looking around, I think DeinoX64 would be a solid replacement for the mpich package atm, and it would simplify what I wanted to do as well as there is only one required package ( be it in x86 and x64 variations.. and even while Mpich is easier to set up since Deino does not allow hooking ).

Idk actually, maybe it's not that important, maybe they already have a plan for when the clients are matured enough. My opinion is that the clients are Stanfords biggest asset, but the way they are managed their biggest drawback. When I look at boinc, that's what I want for F@H aswell. One interface to control all the clients capable of running on the hardware.

Edit: just an example, deino is said to be more stable, yet only came in x86 flavour. But why can I now run DeinoX64 and the folding client without issues? Isn't it easier if there where only one required library?

[Ivoshiee]

Point taken "Ivoshiee", but how many "Outsiders" are there? And how can you "apply"?

It depends how "out" you intend to be. I know just one currently active volunteer. It is relatively easy to apply for that position - just ask for it. Newcomers will likely turned down and directed away from the actual FAH client itself (like tackle with GROMACS code), but if you have been around for years and have some FAH related 3rd-party applications already available then all things are possible. At the end it is up to Prof. Vijay Pande to decide who he will let close to the FAH client code.

[VijayPande]

The main issue is that many people want to help, but when they realize the complexities of FAH, they end up not helping. This drains time from our dev team since it requires coordination to get them code, etc. So, we work with people who are clearly determined and capable of making a contribution.

[v00d00]

But surely there exists people who could handle the menial functions of the client. The non bio-physics part, if you will. People who could extend the functionality of the backend, the actual FAH client, while leaving you and your team to work on the FahCore part.

I think that would be useful, giving us a fuller client with more if you will, eye candy, built in stats tracking, and other useful things.

Maybe something similar to the google summer of code concept.

[MtM]

But surely there exists people who could handle the menial functions of the client. The non bio-physics part, if you will. People who could extend the functionality of the backend, the actual FAH client, while leaving you and your team to work on the FahCore part.

I think that would be useful, giving us a fuller client with more if you will, eye candy, built in stats tracking, and other useful things.

Maybe something similar to the google summer of code concept.

That's exactly what I want to

I just got reminded again how important it is to make my effort open source. I think I just have to have trust in that process, as it will hopefully lead to more support

[7im]

The Amber and Gromacs fahcores are based on open source projects. Anyone who wished to help speed up processing of work units could happily donate their programming or testing talents to those open source projects. But as Vijay said, very few people take up the challenge.

Even the client viewer (i.e. video playground thread) was opened up on this forum for volunteers, and no one came forward yet.

As for extending the client functionality, those who have demonstrated their programming ability to do so with 3rd party utilities are the people who might be asked for assistance. And those who actively participate in a positive manner in this forum, and beta test, might have an extra bullet point on their qualifications for consideration. Those are the people that as Vijay puts it, are "clearly determined and capable of making a contribution."

I'd recommend starting with the Open Source FAQ: http://folding.stanford.edu/English/FAQ-OpenSource

[VijayPande]

But surely there exists people who could handle the menial functions of the client. The non bio-physics part, if you will. People who could extend the functionality of the backend, the actual FAH client, while leaving you and your team to work on the FahCore part.

I think that would be useful, giving us a fuller client with more if you will, eye candy, built in stats tracking, and other useful things.

Maybe something similar to the google summer of code concept.

That's exactly what I want to

I just got reminded again how important it is to make my effort open source. I think I just have to have trust in that process, as it will hopefully lead to more support

We're working closely with Uncle Fungus right now on improving the client. I can ask him what he thinks about adding more people. The coordination issue becomes important.

[codysluder]

But surely there exists people who could handle the menial functions of the client. The non bio-physics part, if you will. People who could extend the functionality of the backend, the actual FAH client, while leaving you and your team to work on the FahCore part.

Let's be realistic. The development delays are probably 99% in the biophysics/FahCore part or in the server code itself. There are almost no delays associated with JUST in the client, and uncle_fungus seems to be on top of all of them. I'm talking specifically about MPI (which is a critical part of FahCore_a1 and _a2, not the client) and about the development of a reliable GPU core.

A reliable viewer is the only thing that MIGHT be open-sourced, but since the API is already open, you can write a replacement if that's your thing. Moreover the PG doesn't have to waste their time sorting out good suggested fixes from bad ones.

[v00d00]

Dont mind me Cody, i have no intention personally of extending the client. It does for me what i need at present.

But if someone else wants to add some sort of extended stats facility, or some other useful functionality, then some sort of Plugin API should be available to allow you to do so.

The viewer was a good start, but as you mention, no one bothered with it, mainly because seeing loads of stuff on screen is stealing cycles from the folding task. I dont think it was something people needed. Also the people who requested that feature, didnt exactly strike me as the programming type. Now if something similar to FAHMon was integrated directly into the client, then i guess you might see a lot more interested people.

Im happy to let Ungle_Fungus help out though as he is one of only a few i would really trust with things, and he already proved himself with the forum.