sub reduce_to_first_sentence {
  local $_ = shift;
  s/([.!?])\s.*/$1/s;
  $_;
}
[download]

-- Randal L. Schwartz, Perl hacker

except my example truncates @ any number of starting characters, not just the first sentence.

($_[0]=~/^(.+[!?.]\s)/s)&&$1||$_[0]
[download]

I don't know where this idiom started, but please do not propagate it.

-- Randal L. Schwartz, Perl hacker

/me kicks himself.

You're right, though for a different reason. It seems to me that, in this particular case, YY ($1) cannot be false, as it contains at least one character, then one of !?., then a whitespace character. Which cannot be false. Or, if the match fails, it contains nothing and therefore is false as it is supposed to be. Usually.

However, it might contain the $1 from a previous match, which makes you right. Again.

You're right that it probably would not have made any difference in this particular case.

But that's the problem. I keep seeing that meme pop up, being used by people who don't understand that it is not "if-then-else" but something that acts a bit like it most of the time.

So, I can't let one like that go by without comment. Since ?: is actually shorter to type, I'd greatly prefer that this other ugly thing be eliminated and the proper operator introduced at every opportunity.

-- Randal L. Schwartz, Perl hacker

I was looking at this last night and came to pretty much the same conclusion... the potential for a nasty bug is there (hence the 'XX && YY || ZZ' construct should be avoided), but in this specific case I don't see it happening.

However, it might contain the $1 from a previous match...

In that case, the regex failed to match, so you never evaluate the $1 anyway. (i.e. XX is false, so even if YY contains something unexpected, we never actually get tripped up by it)

-Blake