bug#16777: [PATCH] Revert "grep: DFA now uses rational ranges in unibyte

[Paul Eggert]

On 02/21/2014 03:41 AM, Aharon Robbins wrote:
> Grep has general RRI problems.  I think the grep team needs to decide if
> RRI is important for grep or not, and if so whether it is more important
> than collating sequences and equivalence characters.

No kidding it has problems!  It seems like almost everywhere I look 
there's a bug waiting to happen.  For example thanks to you I found this 
bug with grep 2.18:

LC_ALL=C grep '/[a-[.e.]]/'
grep: Invalid range end

Which is bogus, as 'e' is a perfectly valid range end.  But at least 
this is an improvement over grep 2.16 and 2.17, which both dumped core 
on this example!

I prefer RRI myself, and would like to see GNU grep move in that 
direction.  Not sure how to get there, though.