From: Ammar Ali <ammarabuali@...>
Date: 2010-10-23T16:54:42+09:00
Subject: Re: Regexp Unicode property names strange behavior?

--001636c5bd5f35f338049344129f
Content-Type: text/plain; charset=UTF-8

On Sat, Oct 23, 2010 at 9:40 AM, Ammar Ali <ammarabuali@gmail.com> wrote:

> On 1.9.2, I'm seeing an "invalid character property name" error from Regexp
> for the named properties Any, Ascii, and Xdigit, but none of the others. If
> I add the u option to the expression, it works.
>
>
> # All good, with or without u option
> [ 'Alnum', 'Alpha', 'Blank', 'Cntrl', 'Digit', 'Graph', 'Lower',
>  'Print', 'Punct', 'Space', 'Upper', 'Word'
> ].each {|name| puts /\p{#{name}}/ }
>
>
> # Errors raised without u option
> ['Any', 'Ascii', 'Xdigit'].each {|name| puts /\p{#{name}}/ }
>
>
> # Now it's good
> ['Any', 'Ascii', 'Xdigit'].each {|name| puts /\p{#{name}}/u }
>
>
> I expected that all the names would either require the u option, or they
> wouldn't. If it was just Any and Ascii, I would accept it and move on, but
> Xdigit doesn't seem to belong with the other two.
>
> Trying to understand why Any, Ascii, and Xdigit are "special". Any clues
> greatly appreciated.
>
>
> Thanks,
> Ammar
>


It was bad documentation! Xdigit should be XDigit, and Ascii should be
ASCII. Any requires encoding to be specified, which makes sense.

Sorry about the noise.
Ammar

--001636c5bd5f35f338049344129f--