1. Many options, but basically you must account for the difference between substrings and stand-alone strings, and this appears to be whitespace in your example data, hence something like:
   s/(?<=\s)-([bc]\s)/-:$1/g;
2. You can stop split from continuing using the 3rd argument:
   split( '\s', 'a b c', 2);
   though if there may be leading whitespace before the 'a' you'd better use the special case of splitting on a single space, which will discard any leading whitespace:
   split( ' ', '  a b c', 2);
   

--
I'd like to be able to assign to an luser

s/-([bc])/-:$1/g; actually doesn't look for the whole string at all--[bc] is a character class, not a literal string.

This means that your regex translates to

• Find a dash ("-")
• Followed by either a "b" or a "c" (save this character in $1)

when you wanted that second line to read "the string 'bc' (save this string in $1)".

The easy solution is to take those brackets out:

and for what I'm guessing is the next step (multiple strings that could match), you can use

But in fact you might be better off with tachyon's suggestion:

The longer but more rewarding solution is to take another couple of trips through perlre when you have a moment--regexen are not the fastest thing in the world to pick up (for us mere mortals, anyway) but they reward study very nicely. :-)

Your other question just needs the last argument to split:

($first, $second) = split ' ', "a  b c", 2;
[download]

Update: fixed boneheaded mistake in split (thanks, Hofmator!)