Re: [RFA 1/3] Linespec rewrite: Parsing

[Keith Seitz <keiths@redhat.com>]

On 03/26/2012 12:31 PM, Tom Tromey wrote:
>>>>>> "Keith" == Keith Seitz<keiths@redhat.com>  writes:
>
> Keith>  +/* A convenience macro for trimming trailing whitespace
> Keith>  +   from string P.  */
>
> We have remove_trailing_whitespace now, see cli-utils.h.

[cough] I even wrote that... I've changed to using it.

> Keith>  +      *(_p + 1) = '\0';					\
>
> I'm curious if this is writing into temporary strings or into the input
> string.

This is called from copy_token_string. The string is allocated via 
savestring, so it should be safe to modify the string like this.

> I ask because I think the module would generally be friendlier if it
> were const-correct and did not write into the input strings.  This would
> make its behavior more predictable.  However I didn't audit it to see
> whether this is reasonably practical; and the language parsers may throw
> a wrench into this idea.

This was a goal. The input stream is never modified. Only a pointer to 
it is advanced, just like we do today.

> I only skimmed the rest, since:
>
> Keith>  I have pushed a patch to my archer branch which permits
> Keith>  quote-encapsulating the linespec. I have also reverted the
> Keith>  "invalid" linespec tests.
>
> ... I think it is time to send fresh patches.

Will do. [And I will follow Pedro's advice.]

Keith