OFFICIAL "SERVER PROBLEMS" THREAD

[paulunderwood]

I have a file with an MD5sum mismatch, What's the remedy?

Edit: The answer was to re-transfer the proof file across my sneakernet,

[S485122]

First of all, thanks for all your work trying to fulfil all the whims of us PimeNet users.

I just looked at the new PrimeNet Activity Summary page dated 2022-12-04 08:00 UTC and have a few remarks.

The number PP1 and PM1-S assignments are not showed in the lower ranges although they make the exponent unavailable.

Still in the lower ranges the number in the TF/ECM column counts a number of assignments, not a number of exponents. But it is not important since those ECM assignments do not make the exponent unavailable.

For each range the numbers of primes should be equal to the sum of the number of Mersenne Primes, factored, verified, unverified, ERR and unchecked. But it seems that on each line where ERR is not 0 the sum is that same number too small (I can't directly point out from which column the number is subtracted while it shouldn't, I suppose it is the number of unverified exponents but it could be from the number of unchecked exponents.)

On the active Assignments page, the PP1 type of work is not selectable and will thus always be shown.

[James Heinrich]

Originally Posted by S485122

On the active Assignments page, the PP1 type of work is not selectable and will thus always be shown.

Added the appropriate checkbox.

[S485122]

Originally Posted by James Heinrich

Added the appropriate checkbox.

Merci !

[Madpoo]

Originally Posted by S485122

First of all, thanks for all your work trying to fulfil all the whims of us PimeNet users.

I just looked at the new PrimeNet Activity Summary page dated 2022-12-04 08:00 UTC and have a few remarks.

The number PP1 and PM1-S assignments are not showed in the lower ranges although they make the exponent unavailable.

Still in the lower ranges the number in the TF/ECM column counts a number of assignments, not a number of exponents. But it is not important since those ECM assignments do not make the exponent unavailable.

For each range the numbers of primes should be equal to the sum of the number of Mersenne Primes, factored, verified, unverified, ERR and unchecked. But it seems that on each line where ERR is not 0 the sum is that same number too small (I can't directly point out from which column the number is subtracted while it shouldn't, I suppose it is the number of unverified exponents but it could be from the number of unchecked exponents.)

On the active Assignments page, the PP1 type of work is not selectable and will thus always be shown.

I'll have to check on that. The new format pulls the same data, just displays it different. The values of each of those is generated separately and probably hasn't been updated to include P+1, etc. I don't remember how long ago the P+1 work became available for assignments, but we probably just need to update the reporting side to account for those.

If you're able to point to a specific exponent range where the totals didn't match what you were expecting, that would be great. The source data was "interesting" and the previous use of a SQL cursor to work out the totals was a little convoluted, so when I rewrote that I had to try and figure out exactly how it's calculating some of those values that aren't directly stored. For example, the source data doesn't directly have how many "untested" exponents are in a range, so that's calculated by subtracting from the total all of the LL/PRP, factors, primes, etc. I think I got it all right but I may have missed the math somewhere.

There was also an interesting "bug" where the "untested" was showing a negative value in a couple of exponent ranges. Turns out that was because a few exponents had a PRP test done on them but later on a factor was found but then the PRP test wasn't marked properly as "PRP, but factor found". That exponent was being counted twice due to the logic involved, and it just took some hunting to find those results and fix them.

That was actually a problem on the old format but because of the way it generated the numbers, the mismatch was being hidden... it wouldn't show a negative value for anything, but if anyone had checked, the numbers wouldn't have totaled up correctly. I think it was in the 4M and 5M exponent ranges. If anyone has that page saved from before a couple days ago, you could check those ranges and see what I mean.

For example, on the 4M range, there's a total of 65,367 exponents, but if you had looked at the "factored" and "verified" column and added them, it would have been 65,368... one too many. :) Because one exponent was being counted in both of those categories by mistake.

[Madpoo]

Originally Posted by chalsall

Thanks for that knowledge.

Interesting that there isn't a date stamp anywhere...

Coulda. Shoulda. Woulda. Manage now.

When the data generates again at the top of the hour, there will be a line at the bottom of that text output that looks like:

Report Timestamp : 2022-12-04 19:00

I added it at the bottom so it would have less impact on any parsing tools anyone has. Adding at the top may have broken something, and I've done enough of that already. LOL

[chalsall]

Originally Posted by Madpoo

I added it at the bottom so it would have less impact on any parsing tools anyone has. Adding at the top may have broken something, and I've done enough of that already. LOL

LOL. And thanks!

[Madpoo]

Originally Posted by Madpoo

I'll have to check on that. The new format pulls the same data, just displays it different. The values of each of those is generated separately and probably hasn't been updated to include P+1, etc. I don't remember how long ago the P+1 work became available for assignments, but we probably just need to update the reporting side to account for those...

I updated the stats so that for assigned work, it will now include P+1 and P-1 (small) into the P-1 category. And I changed the header row to be P±1 to reflect that.

Note that I accidentally changed the header on the "available" category for P-1 as well even though there's really no P+1 "available" work to worry about. I fixed that so when it generates again at the next hour it'll just show "P-1" for that header. Just a cosmetic thing until then.

EDIT: Never mind, I manually fixed those headers now so they're correct.

[Madpoo]

Originally Posted by chalsall

LOL. And thanks!

Sure. And because you asked, on the HTML version I added a blank row every 10M exponent range. That should show up when it generates again in 8 more minutes.

[chalsall]

Originally Posted by Mark Rose

cat foo.json | jq . -r --output csv

🤫

LOL... Does that even work at the command prompt under Wincrows 11? 9^)

[chalsall]

Originally Posted by Madpoo

That should show up when it generates again in 8 more minutes.

Approaching 00:00 UTC. I will click the reload button at 00:02, just to give your code some time to think... 8^)