If you are using Subversion, stop it. Just stop. Subversion = Leeches. Mercurial and Git = Antibiotics. We have better technology now.

Geez, I just barely got started with Subversion and now it’s time to switch already? Seriously, the distributed version control model requires a different way of looking at versioning and branching, which Joel brings into clear relief. There’s also a link to a nice tutorial as well.

Txr is a baroque and painfully hard to use language inspired by, among others, the idea of reversing “here document” generation into “here template” extraction. Since its inception in September 2009, it has grown hair, such as functions that aren’t really like normal functions, and try/catch/finally exception handling. If a complicated Txr query fails on your sample input, just give up. Don’t even think about trying to understand the debug trace output, and the mailing list is likely to be of little help, since pretty much only the author reads it. It is recommended for those who are faced with some simple, boring little problem that is dire need of compounding.

If only all software package descriptions were so honest.

I’m sure that the IT managers meetings don’t resemble this at all.

A couple of you may recall a presentation at UNH back in 2008 on this exact subject (synopsis of presentation) by one of Sun’s GNOME developers. That was clearly some good work those people were up to.

One of the criticisms of Sun, as a for-profit company, was that it had become more focused on values than the bottom line. Maybe so. Perhaps this is why they became Oracle’s dinner.

But as an Oracle customer, UNH should be concerned that the company’s actions are out of step with our core values of diversity and inclusion. We *know* that making allowances for each and every member of our community consumes extra resources.

We should tell Oracle that we see that as a challenge, not a waste.

That’s from a neat little post by Jeffrey Kegler called “Ringo Starr and Willy Sutton On Programming Languages“. Sums up my thoughts on why so much commercial software sucks despite the price tag.

• This one actually affected us at UNH mildly: One of the heuristic rules used by SpamAssassin is to watch for messages dated “grossly in the future”, presumably caused by spammers who are careless with such things. But it turned out that the developers’ idea of “grossly in the future” was hard-coded back, well, grossly in the past.  The actual code snippet:
  

  ##{ FH_DATE_PAST_20XX
  header   FH_DATE_PAST_20XX      Date =~ /20[1-9][0-9]/ [if-unset: 2006]
  describe FH_DATE_PAST_20XX      The date is grossly in the future.
  ##} FH_DATE_PAST_20XX
  

  Darn, they really should have caught that before 2010 rolled around. As it was, the normal update of SpamAssassin rules early on January 2 changed the regex to /20/[2-9][0-9]/. Still a number of messages delivered in the meantime got an extra unwarranted bump in their spam scores. Hopefully, SpamAssassin will come up with a better fix sometime before 2020.

  But SpamAssassin is free software; we know we’d never see that kind of blunder in commercial software, right?

• Well, this one is pretty good too: both SMS-reading software in Windows Mobile and some point-of-sale terminals in Australia jumped ahead to the year 2016 on January 1. The bug is apparently caused by timestamps where the one-byte year field is supposed to be interpreted as binary-coded decimal (so 0x10 means “ten”); instead the software did the binary conversion (where 0x10 means “sixteen”). Voila, you’re in the future!

Thank goodness we’d never make such mistakes ourselves.