Python 3 with Pymysql works pretty well, in my experience.


James Hare
President, Wikimedia DC
http://wikimediadc.org
@wikimediadc

On Sunday, May 31, 2015 at 8:27 AM, Dan Andreescu wrote:

Thanks a lot, this indeed seems to have been the root of all problems. Wikimetrics now works like a charm, not a single error this time. On a wider note, many American-designed pieces of software have a problem with non-Latin 1 characters (for instance both Picasa and Content Translation had Alt+s shortcut reserved for saving, whereas it’s the only way most of us write ą, one of the letters of the Polish alphabet). I wouldn’t have guessed that the problem here might be with the very existence of non-Latin 1 signs anywhere in the data. Thanks for spotting it – and for your assistance.
Cheers

This is actually not as American-centric as would seem on the surface.  We really worked very *very* hard to design Wikimetrics to account for any unicode characters.  It's just that Python 2 is *the worst* language in the world at dealing with this problem (except maybe assembly?)  And sadly the mysql python driver seems determined to never support Python 3.

As the guy who chose these libraries unaware of the terrible unicode handling, I want to take this opportunity to apologize :)  And to say that we're not ignoring unicode, it's just complicated. 
_______________________________________________
Wikimetrics mailing list
Wikimetrics@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wikimetrics