Dave's exactly right! So, I'll recommend and explain something which combines our methods ...

I favor one-liners, but that's just a style preference. I changed my \w to [:alpha:], my ^ to \A, and my $ to \z per davido's recommendation, but I stuck with the \d because it's shorter and still works in this case. I left off the m at the beginning because that's the default anyway. I left off the x at the end because this regex doesn't use extended patterns (ref: Extended Patterns).

• \A : Beginning of the string
• [:alpha:] : One alphabetic character (ref: perlre
• [[:alpha:]\d-]* : 0 or more characters consisting of alphabetic characters, digits, or dashes
• \z : The end of the string

--
-- GhodMode

Your solution is still wrong as proposed:

/^[:alpha:][[:alpha:]\d-]*\z/
[download]

The first mistake is that [:alpha:], by itself isn't a character class. It won't compile as you've written it. It needs to be presented like this:

[[:alpha:]]
^_________^______Note the outer set of brackets.
[download]

The next problem is that the OP stated that the first character cannot be a hyphen character, but he didn't exclude numeric digits. Your solution will fail if the first character is a numeric digit.

Also, '^' matches at the beginning of a "line". That probably isn't an issue for this particular regexp, but it is worth noting that '^' is different from '\A', which matches at the beginning of a string.

If you really despise the /x modifier, and prefer to avoid [:digit:] for whatever reason (maybe less typing?), you could rewrite my original solution like this:

m/\A[[:alpha:]\d][[:alpha:]\d-]*\z/
[download]

But I think the version with the /x modifier is easier to read since it keeps individual anchors together, and everything else separate.




Dave