[QUOTE=Madpoo;441143]Brilliant rant![/QUOTE]
Found the time to watch it. [STRIKE]To set the things straight and show my SIWOTI syndrome, what he says about Romanian (having different plurals for groups larger than 20) is false. Actually plurals are kind of easy in Romanian, other things are more complicate. And also partially what he says about other languages (Polish) is false. But we got the point.[/STRIKE]

Edit: Whoops! Scratch that! He is right. Only is not the plural itself that changes, but the phrase changes indeed! I had to think it over. Absurdistan... That is how we call our country sometimes. Not only because if something absurd happens in the world there is a big chance that it happens in Romania, but also, not long ago, thare was a fashion in the world, when you talk about totally absurd things or news and don't know where to place them, you say they are happening in Romania. See for example the vampires stories, as the most common example. That was generally because not much went out about the country, and people outside had no idea what's happening inside. And the name sounded good, resonant, hehe. When I listened to the clip first time I thought that he is doing the same. But after thinking to it, he is totally right! What a shitty language, hahaha....

Server maintenance starting soon
 

It's time for some server updates... the last time was a while ago and in my reckoning, it's due to get caught up (last time was back in February).

Starting in about 10 minutes from now @ 3:10 PM UTC on 2016-09-14, the server will get some patches applied and rebooted. Based on prior experiences, it'll probably need another round of additional patches and another reboot after that, so I'm estimating 30 minutes total.

EDIT: I split the work up a bit so there's a stretch of time here where Primenet is up. The delays are because the server itself is a virtual server, but I'm also updating the host OS itself which adds extra time to the process. I'm finishing up round two of the Primenet server updates and then another reboot and we should be good for another 7 months. :smile:

error code 11
 

Today I submitted some manual results and I got an error on a few exponents where a factor has been found, like this one:
[QUOTE]Error code: 11, error text: Unable to call factorize on 1028042888169828421903 for M895403947[/QUOTE]Submitting the lines again generated no error, everything was fine.
As I did the submission via 2-3 submission forms in parallel I assume that this situation triggered the behavior - maybe a resource needed for checking the factor was busy. (?)
I thought this might be worth mentioning...

[QUOTE=ramgeis;442542]Today I submitted some manual results and I got an error on a few exponents where a factor has been found, like this one:
Submitting the lines again generated no error, everything was fine.
As I did the submission via 2-3 submission forms in parallel I assume that this situation triggered the behavior - maybe a resource needed for checking the factor was busy. (?)
I thought this might be worth mentioning...[/QUOTE]

Did you happen to try submitting during the time period when I was installing updates on the server? I wonder if that could have done something.

When checking in factor, it spawns an instance of yafu to verify the factor (and make sure it's a prime factor). There's really nothing that would prevent multiple copies running at once so it may have been tied to other system resources being tied up as a result of my maintenance work...

[QUOTE=Madpoo;442556]Did you happen to try submitting during the time period when I was installing updates on the server? I wonder if that could have done something[/QUOTE]

It was roughly 10 hours ago.

[QUOTE=Madpoo;441143]Anyway... to keep it on topic, so yeah, I guess I should find a way and time to alter some columns to nvarchar just to make sure we're storing display names (at the very least) in a suitable way.[/QUOTE]

I took a systematic look at the non-ASCII usernames in the database, using a Python script to semi-automate deciphering the [URL="https://en.wikipedia.org/wiki/Mojibake"]mojibake[/URL].

There are about 1920 names that contain at least one character outside the 7-bit ASCII range (up to U+7F). Nearly all of these are from the username field, however in a few cases they are from the computer name field.

Since this data set is derived from the [URL="http://www.mersenne.org/report_ll/"]LL results report[/URL], it only includes users who have returned at least one LL or DC result. Note the [URL="http://www.mersenne.org/report_factors/"]Factors Found report[/URL] does not include user names. There are probably many times more users registered who never returned any result.

Of the approximately 1920 names, about 1480 of them are correctly interpreted as the Latin-1 subset of Unicode (characters up to U+FF), and therefore already display correctly.

The remainder of approximately 440 are a mix of Eastern European, Chinese, Korean, Russian, Japanese, Turkish, Baltic, etc. There is also some bogus UTF-8 on top of Unicode (e.g., João Paulo instead of João Paulo), and some bogus HTML entities on top of Unicode (e.g,, Vladan Vidaković instead of Vladan Vidaković).

In nearly every single case it was possible to identify the correct Unicode name with a very high degree of confidence. There are only a very small number of cases that are ambiguous, for instance:

Text strings that aren't anyone's actual name, such as:
* tr?ýýýá
* òòòòòòò
* ^zu¨¹n
* -n+-¦<U+98> (a six-byte string whose last byte is the unprintable character U+98)

There is also:
* ®êïñï
which really seems like a legitimate string of some kind but I couldn't decode it.

Someone set their username to &#63743; which currently displays literally ([URL="http://www.mersenne.org/report_exponent/?exp_lo=28633351&full=1"]M28633351[/URL]), but if unescaped would turn into the Unicode character U+F8FF, which is in the private use area and is used by Apple to store their logo.

Basically, if you ever do create nvarchar columns, the data to fill them has already been generated.

[QUOTE=GP2;442826]
There is also:
* ®êïñï
which really seems like a legitimate string of some kind but I couldn't decode it.
[/QUOTE]

I think it could be Reini, as in, it's one of those display names where the user is picking symbols that look like the letters they want to use.

[QUOTE=0PolarBearsHere;442885]I think it could be Reini, as in, it's one of those display names where the user is picking symbols that look like the letters they want to use.[/QUOTE]

Hmmm, that does seem to be a Finnish last name and a German first name.

Bug: can't updated account login username in settings.

@George/Aaron: I'm not sure if this is by design or not. Should users be able to [i]edit[/i] their username after creation or not? The same form is used to create and edit accounts, and creating a new account (line 146) uses the user_login field, but when updating an existing account (line 174) the field is not updated.
It's an easy fix, but the question is which fix?
1) Allow users to change their username
2) Make the username field read-only on the form to make it clear it's not editable once created.

[QUOTE=James Heinrich;444976]Should users be able to [i]edit[/i] their username after creation or not?[/QUOTE]Why not? If the DB uses foreign keys properly then it shouldn't be an issue at all.