PRPNet 5.5 Released

[rogue]

I made a change to LLRProgram.cpp to fix the problem. If you build it on your own get the latest code and build it otherwise wait until I can post one.

What happened was that the output from llr changed so that the PRPNet client was unable to correctly parse the version number. The change should be fully backward compatible with older versions of LLR.

[rebirther]

I have built 32/64bit for windows. You can download it here

[rogue]

I have fixed some issues related to generic prime searches.

[henryzz]

I am currently double checking a load of tests that were done before I upgraded my bios and made my pc stable. How should I expect to discover any mismatches? Will they appear in all.html? prpserver.log? I do know that there are a couple of fake primes that were discovered which are bad tests.

[rogue]

Originally Posted by henryzz

I am currently double checking a load of tests that were done before I upgraded my bios and made my pc stable. How should I expect to discover any mismatches? Will they appear in all.html? prpserver.log? I do know that there are a couple of fake primes that were discovered which are bad tests.

Presuming you have your old data in the database, you can compare the residues in the CandidateTestResult table. If you have refreshed your database, then after stripping of the timestamp in completed_tests.log, compare the two files.

[henryzz]

Originally Posted by rogue

Presuming you have your old data in the database, you can compare the residues in the CandidateTestResult table. If you have refreshed your database, then after stripping of the timestamp in completed_tests.log, compare the two files.

So it is a manual process? Does it do third checks itself automatically?
It is the same database. I just turned double checking on in the server.

[henryzz]

Originally Posted by rogue

Presuming you have your old data in the database, you can compare the residues in the CandidateTestResult table. If you have refreshed your database, then after stripping of the timestamp in completed_tests.log, compare the two files.

SELECT a.CandidateName, a.Residue, b.Residue
FROM CandidateTestResult a
JOIN CandidateTestResult b ON b.CandidateName = a.CandidateName AND b.Residue != a.Residue
WHERE a.TestID < b.TestID;

This code returned two mismatches so far.
It seems that unless you actually check the database prpnet gives no indication that there are mismatching residues. I would have thought that this should be in both the log and the webpages. It should be obvious when mismatches occur.

[rogue]

Originally Posted by henryzz

SELECT a.CandidateName, a.Residue, b.Residue
FROM CandidateTestResult a
JOIN CandidateTestResult b ON b.CandidateName = a.CandidateName AND b.Residue != a.Residue
WHERE a.TestID < b.TestID;

This code returned two mismatches so far.
It seems that unless you actually check the database prpnet gives no indication that there are mismatching residues. I would have thought that this should be in both the log and the webpages. It should be obvious when mismatches occur.

The software does not have a page that shows the number of mismatched residues. IIRC, the server stats page has some information on double-checks, but stats on how many candidates required a third test. If you have double-checking turned on on the server, then you can count the number of Candidates records where completedtests > 2 as a mismatch would trigger a third test.

[henryzz]

Originally Posted by rogue

The software does not have a page that shows the number of mismatched residues. IIRC, the server stats page has some information on double-checks, but stats on how many candidates required a third test. If you have double-checking turned on on the server, then you can count the number of Candidates records where completedtests > 2 as a mismatch would trigger a third test.

No third tests have been done yet in the database(looked at table candidate column completedtests and candidatetestresult table). There are two completed tests for the mismatches.
The only reference to double checks I can find on the webpages is Count DC'd. No references to third tests.
Do I have to wait until doublechecks catch up with first time checks?

[rogue]

Originally Posted by henryzz

No third tests have been done yet in the database(looked at table candidate column completedtests and candidatetestresult table). There are two completed tests for the mismatches.
The only reference to double checks I can find on the webpages is Count DC'd. No references to third tests.
Do I have to wait until doublechecks catch up with first time checks?

Yes. If you have other questions or ideas for improvements, please e-mail me.

[henryzz]

Originally Posted by rogue

Yes. If you have other questions or ideas for improvements, please e-mail me.

And next day the triple checks were assigned. Doublechecks are still far behind the first time checks.

Ideally I wouldn't need to use sql to monitor this. This is especially true as it is the end user that needs to know if they are producing incorrect tests. It should be on the webpages.

I have relearnt some sql which is a good thing.