I think of it more like gnutella style. The user can set the percentage/pirority... etc of how much CPU resource consumed by external program, then an resource intensive progam will likely to return the result slower. Since it is NOT possible that all the same hardware platforms are VERY BUSY at the same moment, there will still be some of them return the result very quickly, some very slow. The developer can set the number of sample to be 3, then when 3 sample of each hardware platform is received, a stop command is issued automatically(or he can issue manually) to that group of machine which they will cease working for that program immediately. As long as there will be idle Linux machine or NOT EXTREMELY busy Linux machine, there will be a rich ground of automatical beta-testing. As for the issue of security: That belong to the responsibility of Sharer to create a virtual protected mode, and ANYTHING happen inside must have NOTHING to do outside. It is rather easy to achieve that under Linux. Or we can add a stop button for owner of that computer, if s/he doesn't think what is happening, s/he can manually stop that process. Also an .org may be setup to gave each developer a uniquely identifiable number tag so each progam is tracable to an individual. Yes any program that demand user input is unlikely to be tested in that enviornment because it require VERY LITTLE from the user. The program is likely consist of five compenents: 1. Identifier: indicating the hardware platform of each machine is; 2. Communicator: Send and recieve help,stop signal, program and result to the web; 3. Sharer: Create protected mode of CPU, help managing system resource; 4. Executor: Carry out the testing program 5. CEO: excute EXCO function like stop, start and setting percentage, target hardware platform, number of sample....etc. Once a developer finish writing a program, he use CEO to entre the detail of target hardware platform, along with a simple description of what it does, number of sample, time period, expiry time, then Communicator would broadcast all these to www, the Communicator on the other end receive the signal, CEO check the Identifier if it is the "right" machine, if yes then it init the Sharer with pre-set precentage by the user, the program is then execute by Executor,the result is then send by Communicator along with indicator of hardware platforms. Any problem?
--- Ed Wilts <[EMAIL PROTECTED]> wrote: >On Sun, Jul 20, 2003 at 12:56:04PM -0600, Rodolfo J. Paiz wrote: >> At 7/20/2003 14:51 -0400, you wrote: >> >This world-wide testing environment sounds very good in theory. But what >> >happens of the program >> >being tested is a resource-intensive program? Then my PC would naturally >> >run slower, because someone is testing there program in my protected area >> >of my CPU. >> >> It also only works with programs that require absolutely no interactivity >> and that do not involve any network traffic. > >And only if you can ensure with absolutely certainty that the program >isn't going to do something evil to either your internal network, >system, or to the outside world. And since by definition it has to >communicate its results with the outside world, you're trusting a >developer or the project leader you know nothing about. > >Not on my system you don't. > > .../Ed > >-- >Ed Wilts, Mounds View, MN, USA >mailto:[EMAIL PROTECTED] >Member #1, Red Hat Community Ambassador Program > > >-- >redhat-list mailing list >unsubscribe mailto:[EMAIL PROTECTED] >https://www.redhat.com/mailman/listinfo/redhat-list -- redhat-list mailing list unsubscribe mailto:[EMAIL PROTECTED] https://www.redhat.com/mailman/listinfo/redhat-list
