[GRASS5] g.region G_usage() commented out?

[Glynn Clements]

Russell Nelson wrote:

>  > AFAICT, the code appears to be ensuring that the code processes the
>  > entire string, and not just some initial portion of it.
> 
> But tbuf[] is filled by copying characters from buf up to the first
> null.  Therefore, buf cannot contain any embedded nulls, and
> strlen(buf) is going to be equal to strchr(tbuf, MARKER) - tbuf every
> time.  So why not just run up to the null?  The only reason I can see
> to have MARKER would be if you wanted to safely look at buf[i] and
> buf[i+1] without first checking to see if buf[i] is a '\0'.  The
> original code didn't do that, but ahhhh, I see that I have introduced
> a potential segfault.  Revised patch, which actually is simpler and
> easier to understand:

It still changes the semantics of the function. The existing version
will reject strings which contain trailing whitespace, due to:

    if (h[1] != MARKER)
	return 0;

If the marker is removed, G_strip() will strip any trailing whitespace
before the test is performed.

I'm not suggesting that I have a personal preference for the existing
behaviour; I'm just asking you (or, for that matter, anyone else)
whether you are sure that it's safe to change it.

My *suspicion* is that it *probably* is safe.

-- 
Glynn Clements <glynn.clements at virgin.net>