efficient regular expression that never matches

efficient regular expression that never matches

[cinsk]

hi,

Is there any special regular expression string that never matches
anything?

I'm thinking of "$a", but it should scan all the string until end
of the string, so it may not be efficient.

Is there any short, efficient regular expression that never
matches anything? (it's okay with Emacs-specific regular
expression, since I'm going to use it with `directory-files' or
`string-match' function)
 
Thank you.

Re: efficient regular expression that never matches

[Stefan Monnier]

> I'm thinking of "$a", but it should scan all the string until end
> of the string, so it may not be efficient.

IIRC I used something like "\\`\\'a".


        Stefan

Re: efficient regular expression that never matches

[John Wiegley]

>>>>> Stefan Monnier <monnier@iro.umontreal.ca> writes:

>> I'm thinking of "$a", but it should scan all the string until end of the
>> string, so it may not be efficient.

> IIRC I used something like "\\`\\'a".

Yes, also because "$" only has special meaning as the last character of the
regexp, whereas \\' should always mean "end of <context>".

John

Re: efficient regular expression that never matches

[cinsk]

On Saturday, November 21, 2015 at 12:59:51 AM UTC+9, John Wiegley wrote:
> >>>>> Stefan Monnier <monnier at iro.umontreal.ca> writes:
> 
> >> I'm thinking of "$a", but it should scan all the string until end of the
> >> string, so it may not be efficient.
> 
> > IIRC I used something like "\\`\\'a".
> 
> Yes, also because "$" only has special meaning as the last character of the
> regexp, whereas \\' should always mean "end of <context>".
> 
> John

Yes, I agree with that,

However, "$a" means that the regular expression engine may read all strings until the end of the line.  So, if I can do something like "^<SOME CHARACTER OR CONDITION IMPOSSIBLE>", then the engine should detect before reading all characters of the line, I presume.  Perhaps "^\uFFFFF" can be a solution?  Since \uFFFF is noncharacter of the unicode (\uFFFF can be use for the internal purpose but generally not used.)

Re: efficient regular expression that never matches

[Stefan Monnier]

> until the end of the line.  So, if I can do something like "^<SOME CHARACTER
> OR CONDITION IMPOSSIBLE>", then the engine should detect before reading all
> characters of the line, I presume.  Perhaps "^\uFFFFF" can be a solution?

That's what the \\` is for in my solution.


        Stefan