Re: [bug-grep] Re: grep: -i option not working i cronjobs

[Aharon Robbins]

Aharon Robbins <address@hidden> writes:
>>> My problem with REG_ICASE is that it's a regex-compile-time bit. I'd
>>> rather it be a regex-execute-time bit, since in gawk the use of
>>> IGNORECASE can be changed around dynamically.

Andreas:
> > That's not a real problem.  You can compile both forms and use the
> > appropriate one for matching depending on the current value of 
> > IGNORECASE.

I could.  And for FS and RS, I do.  But I'd prefer to avoid both the
space overhead and the execution time overhead of doing that for every
possible use of a regular expression.  At the moment, gawk just
dynamically recompiles if IGNORECASE has changed since the last time
a regexp was compiled.  I suppose that one can find pathological cases
to demonstrate that both ways are expensive.  Thus my preference to
try and use a "somebody else's problem field." :-)

From: Elliott Hughes <address@hidden>
> (reads GNU awk user's guide.) ah, i see. i didn't know about that. i'm 
> not sure i ever used GNU awk. i thought he meant support for Perl-like 
> "(?i)" in regular expressions.

Nope.  I tend to doubt that perl-style RE's and regular old boring POSIX
style RE's should mix.  But maybe I'm just too conservative.

Thanks,

Arnold