On 11/9/09, David Anderson <[email protected]> wrote: > The goal is that a BOINC server with the default settings should > address these issues in a way appropriate to a production project: > - validation > - credit granting > - job runtime estimation > - assignment of deadlines > > Of course, everything can be overridden, > but this should be the exception rather than the rule. > Expecting project admins to understand a bunch of complex > issues and sift through mountains of configuration options > has, I think, significantly limited the adoption of BOINC by scientists.
I think lack of tutorial-like documentation is what causes problems with adoption, not lack of good defaults. If you make everything Just Work work out of the box with defaults (like installing an assimilator that copies the output file to a 'output_files' directory), then the admin won't learn how some things works, or even that they exist (for example, he could end up being unaware of the existence and purpose of an assimilator). That's why I've always been against --test_app... But we're drifting off-topic already. -- Nicolas _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
