Dangerous thinking in the perl core...

Matts on 2006-10-23T01:47:08

> It doesn't address that I'm trying to get information out of the diagnostic string. > It also doesn't address the far more common case of Perl code producing error > messages to emulate perl's own.

Since perl doesnt make any promises about error messages id say that any code doing this is not exactly entitled to complain when things break. The fact that any given version of perl documents a particular error doesnt mean that we promise to provide that error for ever more in that situation.
I think this is dangerous. For a long time perl didn't make a lot of guarantees -- until the documentation was updated to detail them. But perldiag is right there detailing the error messages, and people do parse for them, every single day, because they have no other choice right now.


Exception Objects

chromatic on 2006-10-23T01:58:50

... which is why Yves, Schwern, and myself have argued over the past few days over providing a better interface to warnings and error messages. Parsing natural language is fragile. There are better ways to make things unambiguous.

Re:Exception Objects

Matts on 2006-10-23T02:14:01

Yeah I got around to reading most of the rest of the thread now (spent the weekend in Hospital after my wife fell on her head while ice skating, so this weekend I've had very little time to catch up on p5p). I just thought that one early post was rather dangerous thinking.

Re:Exception Objects

sigzero on 2006-10-23T13:38:23

Oh man, I hope she is okay.

Re:Exception Objects

Matts on 2006-10-23T13:43:11

I think she'll be fine. They're keeping her in for observation right now, but I'm sure she'll be out soon.