Double checks

[casmith789]

I finished an exponent today (3438827) which matched the DC residue of another test, so it's been completed.

Great, you might say.

But looking at the exponent it seems to have already been tested 4 times with the same result and once with a different result back in 2007. Why was this exponent not already finished before my test?

http://www.mersenne.org/report_expon...4388257&full=1

[firejuggler]

probably because the check was done mutiple timewith the same machine.

[retina]

Originally Posted by casmith789

http://www.mersenne.org/report_expon...4388257&full=1

What makes me curious is why the user "David Marsland" in the LL section shows up as "ANONYMOUS" in the history section?

[casmith789]

http://www.mersenne.org/report_LL/?e...B1=Get+LL+data

This page seems to suggest that only my DC run and James Harris's were run successfully, and there weren't 4 different runs. So why are the histories different?

mersenne.ca didn't shed any light on it either.

[Madpoo]

Originally Posted by retina

What makes me curious is why the user "David Marsland" in the LL section shows up as "ANONYMOUS" in the history section?

The names in the LL results are from the user's "version 4" name. The old log messages in the history section will just show "anonymous" if that user never created and mapped a v5 account to their old v4 account.

The multiples in the history section are because the user submitted the same result multiple times. We do have plans to try and remove those spurious dupes where it was clearly a case of the user re-submitting over and over, so hopefully we'll get these weird things taken care of.

[Gordon]

Originally Posted by Madpoo

The names in the LL results are from the user's "version 4" name. The old log messages in the history section will just show "anonymous" if that user never created and mapped a v5 account to their old v4 account.

Actually that's not true, witness this report 637447 where I most certainly have mapped my V4 results to the V5 database. This isn't the only one either.

[Prime95]

Originally Posted by Gordon

Actually that's not true, witness this report 637447 where I most certainly have mapped my V4 results to the V5 database. This isn't the only one either.

Correct. I believe all the v4 log entries show up as anonymous in the history section. We ought to be able to generate a SQL query that fills in the userid in most cases (v4 ids were case sensitive, v5 ids are not which may cause some difficulties).

So our to v4 log to do list includes:

1) Generate the v5 user id in the history section where possible.
2) Remove v4 duplicate rows, keeping the oldest entry.
3) In the LL results table, fill in the date reported when it can be determined from the v4 logs.

Anything else?

[Madpoo]

Originally Posted by Prime95

Correct. I believe all the v4 log entries show up as anonymous in the history section. We ought to be able to generate a SQL query that fills in the userid in most cases (v4 ids were case sensitive, v5 ids are not which may cause some difficulties).

So our to v4 log to do list includes:

1) Generate the v5 user id in the history section where possible.
2) Remove v4 duplicate rows, keeping the oldest entry.
3) In the LL results table, fill in the date reported when it can be determined from the v4 logs.

Anything else?

FYI for George: when I was finding LL results where the first/second check were done by the same user, but first on V4 and then again with their V5 user, I built a function to map their v4 user name to their v5 user id. It might not be precise enough for step #1 there due to the CI/CS difference, but it was good enough for what I was using it for. It's "fn_legacy_userid_mapping" -- pass in the v4 user id string and you'll get a matching v5 user ID # (if any).

I won't vouch for how well it would perform if it were used in bulk.