June 8th, 2003

binary and a cookie to whoever notices
  • xb95

Synsuck Wide Character Issues

For the last ... forever ... synsuck has been giving the following error and dying:

Wide character in subroutine entry at /home/lj/cgi-bin/ljprotocol.pl line 659.

This code is:

my $dupsig = Digest::MD5::md5_hex(join('', map { $req->{$_} } qw(subject event usejournal security allowmask)));

I've investigated this, and it seems to be caused whenever someone has some Unicode characters in their entry. It hasn't happened too much, every few days it'll happen, but it tends to get bumped out of the feed after a few cycles, and stop being a problem. But as Plogs.net grows, this is going to start becoming more of a problem.

I thought I had seen something in this community regarding this problem, but I looked back 60 odd entries and couldn't find anything. So, any thoughts on how to fix this? Does LiveJournal have this problem? (And if not ... how? AFAIK, this code hasn't changed in quite a while. Although, you guys pull syndicated feeds every 30 minutes, right? We pull every 3 hours. Eh.)