mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   PrimeNet (https://www.mersenneforum.org/forumdisplay.php?f=11)
-   -   Primenet web design (https://www.mersenneforum.org/showthread.php?t=19716)

Madpoo 2017-06-18 19:19

Changes to the expiration info on assignment reports
 
I'm working up some changes to how "days until expiration" is calculated on the assignment pages.

Currently it's using the typical assignment rules for how many days you have to finish them and also handling the special case of category 1 work has to report in at least every 30 days.

What it wasn't doing was taking into account assignments that were being expired due to "unresponsive computer" rules.

Those rules indicate an assignment will expire under these conditions:
It has been 10 days or more past when the computer said it would check in again
AND
It has been 60 days since the computer last checked in at all
(this is actually 40 days for double-checks by a computer that hasn't done any previous work...I'm not sure why but first-time checks don't have this "new computer" lower threshold)

The rules also only apply to exponents in what I've referred to as the "butter zone" although George probably has a more scientific sounding name :smile: For double-checks, the range is 39e6 to 59e6 (it gets updated occasionally to be about 20e6 + whatever the lowest DC exponent is). For first time checks it's currently 70e6 to 110e6 (40e6 + the smallest first-time check)

So if you're testing large exponents (100M digit stuff) and leave your computer unattended for years, disconnected from the Internet, it'll still be fine.

Anyway, as you can imagine, those rules make calculating expirations a little trickier. I ended up with calculating the expiration under the "you have XX days to finish', and also under the "cat 1 need to update within 30 days" and also now these "non-responsive CPU" rules. Then I figure out whichever one is the smallest since that's the expiration it will use.

What's worse, under those non-responsive rules, there are actually two things at work. Has it been over 60 days since the last update, but still under 10 days since it's expected to check in? Or alternately has it been > 10 days since it was supposed to report, but it's less than 60 days since the last update? Well heck, I have to look at both cases.

TL;DR:
From now on, the expiration estimate will do this:
1) If it has been >= 60 days (40 for dc's on a new cpu) since the last update, but not yet 10 days past the date it was next expected, it figures it will expire once it is 10 days past that "next expected" date (assuming that's smaller than any other expiration consideration)
2) If it has been >= 10 days overdue since it was supposed to check in again, and it'll cross 60 days since the last update (or 40) sometime in the next 7 days, the days-to-expire will be that countdown to when it reaches 60 (or 40) days.

Whew. Confusing, I know, but then we are dealing with things that expire for different reasons and trying to estimate it means making a few assumptions...

By way of example, consider the following case:
M41539451 was assigned in Dec 2016, was last updated on the same day (hasn't even started yet), and it was expected to check in again today (2017-06-18).

Under the old expiration estimate it didn't have one... it was originally category 3 and currently category 2, so none of those rules have kicked in yet. But in truth, if it doesn't report back in now in the next days, it will expire on 2017-06-28 so it's actual days-to-expire will show up as 10.
[URL="https://www.mersenne.org/assignments/?exp_lo=41539451&exp_hi=41540237&execm=1&exfirst=1&exp1=1&extf=1"]Example[/URL]

I hope this is all helpful info and will answer any questions about "why is the expiration showing XX for assignment YY"

LaurV 2017-06-22 14:12

Confirming here about FF54 muffling the blue color of the gimps page. FF53 was ok. It may be a FF thing...

tha 2017-06-27 06:46

[QUOTE=LaurV;461772]Confirming here about FF54 muffling the blue color of the gimps page. FF53 was ok. It may be a FF thing...[/QUOTE]

It sure looks like that if FF53 was doing it fine. However FF tends to be less forgiving (browser correcting obvious errors in HTML formatted documents) than others so it might be a good idea to implement some of the recommendations from:

[URL="https://validator.w3.org/check?uri=https%3A%2F%2Fwww.mersenne.org%2Freport_exponent%2F%3Fexp_lo%3D14753149%26full%3D1&charset=%28detect+automatically%29&doctype=Inline&group=0"]https://validator.w3.org/check?uri=https%3A%2F%2Fwww.mersenne.org%2Freport_exponent%2F%3Fexp_lo%3D14753149%26full%3D1&charset=%28detect+automatically%29&doctype=Inline&group=0[/URL]

Gordon 2017-06-27 16:22

[QUOTE=snme2pm1;460400]Ed: Just looked up ISO 9601, but did not find any mention of quarters! Bother![/QUOTE]

Quarter dates are steeped in history (here at least) and were the days on which taxes were collected and rents became due, used since the Middle Ages

25th March
24th June
29th September
25th December

Apart from Christmas day the way to remember is count the letters in the month and add that to 20.

Mark Rose 2017-06-27 18:04

[QUOTE=Gordon;462180]Quarter dates are steeped in history (here at least) and were the days on which taxes were collected and rents became due, used since the Middle Ages

25th March
24th June
29th September
25th December

Apart from Christmas day the way to remember is count the letters in the month and add that to 20.[/QUOTE]

[url]https://en.wikipedia.org/wiki/Quarter_days[/url]


All times are UTC. The time now is 21:28.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.