Hi,
I am reading std.regex. I found the following code on line 671:
        else if('A' <= current && current <= 'Z')
            val = val * 16 + current - 'A' + 10;

Hex digits are parsed there. So unles this is something special, this is a bug. Because for example what 'J' would mean in this context?

14-Feb-2013 22:24, Lubos Pintes пишет:
> Hi,
> I am reading std.regex. I found the following code on line 671:
>          else if('A' <= current && current <= 'Z')
>              val = val * 16 + current - 'A' + 10;
>
> Hex digits are parsed there. So unles this is something special, this is
> a bug. Because for example what 'J' would mean in this context?

A bug. Especially considering the full source:

	if('0' <= current && current <= '9')
            val = val * 16 + current - '0';
        else if('a' <= current && current <= 'f')
            val = val * 16 + current -'a' + 10;
        else if('A' <= current && current <= 'Z')
            val = val * 16 + current - 'A' + 10;

I'd say file it or better yet submit a pull with a unittest that presently  does something fishy. For instance "\u00JJ" would parse and match m-hm 'J'-'A' * 16 + 'J' - 'A' where it shouldn't parse in the first place.

-- 
Dmitry Olshansky

On Thu, Feb 14, 2013 at 07:24:32PM +0100, Lubos Pintes wrote:
> Hi,
> I am reading std.regex. I found the following code on line 671:
>         else if('A' <= current && current <= 'Z')
>             val = val * 16 + current - 'A' + 10;
> 
> Hex digits are parsed there. So unles this is something special, this is a bug. Because for example what 'J' would mean in this context?

Looks like a bug. The last clause of the if-condition should check <= 'F', not <= 'Z'.

Please file a bug at http://d.puremagic.com/issues .


T

-- 
GEEK = Gatherer of Extremely Enlightening Knowledge