Re: CONFIG_KPROBES=y build requires gawk

[Masami Hiramatsu <mhiramat@redhat.com>]

Al Viro wrote:
> On Wed, Dec 16, 2009 at 10:43:47PM -0500, Masami Hiramatsu wrote:
>> Al Viro wrote:
>>> On Wed, Dec 16, 2009 at 03:56:17PM -0800, Andrew Isaacson wrote:
>>>> With CONFIG_KPROBES=y on Ubuntu 9.10 x86_64 default install, I get the
>>>> following:
>>>>
>>>> Error: Your awk doesn't support charactor-class.
>>>> Please try to use gawk.
>>>
>>> Aside of the incorrect suggestion (you need not just gawk, you need that
>>> thing installed as awk), the use of GNUisms in there is actually pointless
>>> since encoding is bloody fixed.
>>
>> Hmm, maybe "POSIX awk" will be a better suggestion, isn't it?
> 
> Not really.
> 
> Note that meaning of [:alpha:] depends on locale, and you are working
> with fixed input, in fixed encoding.  So you *already* have a dependency
> on locale and it's no better or worse way to express the set than A-Za-z
> is.  With LANG=C both describe the set you are after; in other locales
> neither is guaranteed to.  And having it set by caller of your script
> is trivial.
> 
> IOW, you are actually misusing the language feature and while it's easy
> to fix, the same fix will remove any benefits compared to universally
> available alternative language feature.

Ah, I see:)

I think that this is not so big issue as making new language in the kernel.
(wow, if someone wants to do that, I don't dare to stop him :) )
If just typing less than 100 characters can fix it, I'd like to do it; e.g.

Ualpha = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
Lalpha = "abcdefghijklmnopqrstuvwxyz"
alpha = Ualpha Lalpha
digit = "0123456789"
alnum = alpha digit

and replace character classes with it. ("[[:alpha:]]"->"[" alpha "]")

That's so easy, isn't that? :-)

Thank you,

-- 
Masami Hiramatsu

Software Engineer
Hitachi Computer Products (America), Inc.
Software Solutions Division

e-mail: mhiramat@redhat.com