Canonical Result
From Unofficial BOINC Wiki
[edit] General
The simplest and best Result that is chosen from the set of Results that are part of the Quorum of Results and is saved as the example Result in the Science Database (see Assimilation Process).
If there are still Results being processed, and the Result Deadline for those Results has not been passed, the File Deleter can be used to remove the other Result Data Files that may be present to free up disk space. The Canonical Result is thus retained as the "model" for all of the other Results for this Work Unit not yet returned.
[edit] Choosing a Canonical Result
Inside the SETI@Home Validator, for example, it's easy; you'll have, at most, Result #1, #2, #3, #4, #5, #6, although normally there are only 3 Results. Result #1 is checked against #2 to see if they are strongly similar. If the answer is yes, then, Result #1 is the Canonical Result, and both Results are marked as Validated Results. Then Results #3, #4, #5 and #6 are afterwards checked to see if they are weakly similar to Result #1.
If Result #1 and #2 are not strongly similar, Result #1 is compared against #3, #1 against #4 and so on to #1 against #6. If none are "strongly similar", Result #2 is tried against #3 and so on to #5 and #6. Immediately after the Validator finds any two "strongly similar" Results, the remaining tests are aborted and one of the Results is chosen as the Canonical Result. Both are marked as Validated Results. All the other Results are then checked to see if they are weakly similar to the Canonical Result.
But Results may appear to have been chosen randomly as #1, #2 and so on. This order doesn't need to be same order as result-id in db, nor order of sent out, or order of returned, since there are instances the last result-id, last sent out and last returned of the 3 initially is Canonical Result when only 3 returned yet, while in other instances it's 1st id, 1st sent out, 1st returned...
So the only things you can be certain about are:
- The Canonical Result is strongly similar to at least one of the other Results.
- If there is "no consensus yet", no pair of Results was strongly similar.
- Any Result that didn't error-out for various reasons, and is marked "invalid" is not weakly similar to the Canonical Result.
Note, projects other than SETI@Home use their own Validator and may work differently.
