Re: [PATCH 1/3] grep: add --warnings={always,never,auto}.

[Paolo Bonzini]

On 08/31/2010 11:48 AM, Jim Meyering wrote:
> I was wondering why you chose not to make grep exit with status 2
> from the beginning.  There is no legitimate use for the type of
> regular expression that would evoke this failure, so it is unfair
> to compare grep's new warning with gcc's -Werror, which can be
> triggered for valid and useful constructs.

Some of them are _extremely_ doubtful (-Wformat, for example).  But they 
are not errors anyway.

> In line with my personal preferences [if I make the mistake, I want to
> find out as soon as possible], I think we should make grep work as
> Paul suggests.  The value/risk ratio of such a change is unusually high,
> since the likelihood of a valid use triggering an unwarranted failure
> is so low, and the typo of omitting the outer [...] in e.g., "[[:lower:]]"
> is so common.
>
> I don't see what concern could make you object to this change.

Because even though we agree that the regex we warn about is 
(practically) always going to be semantically incorrect, there's no 
doubt that the regex are syntactically valid, and this is IMHO enough 
reason to not return an exit status of 2.  Rewriting it as [[:lower:]] 
would be as good then (or as bad).

Second, if this was done, it should operate in the same way in sed, 
expr, awk, and all other GNU programs that deal with regexes.  (And 
possibly in glibc too).

If you want to add --warn=error (which is a "superset" of --warn=always 
behavior), that's fine and I actually like the idea.  But I think making 
it the default non-POSIXLY_CORRECT behavior is wrong.  Honestly, if this 
happened I would regret having introduced the feature in the first place.

Paolo