Re: [PATCHv2] dprintf implementation

[Vincent Pelletier]

Yoshinori K. Okuji wrote:
> On Friday 25 February 2005 17:02, Hollis Blanchard wrote:
> > int
> > grub_strword (const char *haystack, const char *needle)
> > {
> >     int pos = 0;
> >     int found = 0;
> >
> >     while (haystack[pos]) {
> >         /* Advance to next word. */
> >         while (grub_iswordseparator (haystack[pos]))
> >             pos++;
> >
> >         if (0 == grub_strcmp (&haystack[pos], needle))
> >             {
> >                 found = 1;
> >                 break;
> >             }
> >     }
> >
> >     return found;
> > }

What about grub_strword("all,foo","all") ?
Wouldn't strcmp tell that {'a','l','l',',',...,'\0'} and 
{'a','l','l','\0'} are different ?

> grub_strword ("filesystem", "file") returns 0 in your implementation.

Isn't it what should happen ?
If we use 2 similar words, one being a substring of the other, I don't 
think the substring should make the larger match too. Or we might add a 
hierarchy thing (filesystem/inode, filesystem/block, kern/alloc, 
kern/term/font/draw, ...) where "filesystem" would match all 
"filesystem/*" ad "filesystem/inode" would only match that one.

Hollis: mmh, I bet you'll say it's overcomplicated :). And I think 
you're right. The more I think about such "simple" debug func, the more 
I feel like I'll take over the world of debugging system... Well, time 
to find some more of those nice mushrooms ;). And a clean paper to write 
those 500 Changelog entries.

Vincent Pelletier