Re: [PATCH] Merge forward-search/reverse-search, use gdb::def_vector, remove limit (Re: [RFA] Fix leak in forward-search)

[Pedro Alves <palves@redhat.com>]

On 11/30/2018 08:42 PM, Tom Tromey wrote:
>>>>>> "Pedro" == Pedro Alves <palves@redhat.com> writes:
> 
> Pedro>   -/* FIXME!!!  We walk right off the end of buf if we get a long line!!! */
> 
> I'm surprised all those exclamation marks didn't help us find this
> sooner.  /s
> 
> Pedro> I don't think there's much point in reusing the buffer across
> Pedro> command invocations.
> 
> Agreed.
> 
> Pedro> -  msg = (char *) re_comp (regex);
> Pedro> +  char *msg = (char *) re_comp (regex);
> 
> Pre-existing, but I wonder why this cast is needed.  I think "const char
> *msg" and removing the cast ought to be completely fine.  

Yeah.  I did this, and pushed the patch.

> Or (more work)
> avoid re_comp and use compiled_regex instead.

I looked at this a bit, but decided not to do it, at least not in
this patch, because there are several other re_comp calls in the
tree, which made me think that it'd be better to do a pass
tweaking all the same time.  While considering that, it seems like
we end up having to pass a string as 3rd argument to compile_regex's
ctor and many different places, which makes me think that perhaps
that argument should have a default.  Similar consideration for
the compile_regex::exec method, and all arguments but the first.

Thanks,
Pedro Alves

> 
> The rest seems fine to me.   Thank you for doing this.
> 
> Tom
>