On 26 June, 2015 - Linus Torvalds wrote:
> On Fri, Jun 26, 2015 at 2:50 PM, Anton Lundin <[email protected]> wrote:
> > We store no value as a NULL, so we can omit that value from the xml and
> > git files.
>
> Ack.
>
> However, it strikes me that maybe we should just teach "copy_string()"
> to do this. That would match "same_string()", and copy_string()
> already returns NULL today (although only if you passed in NULL), so
> it really would match that too.
>
> If people wanted strdup(), they should use strdup(). copy_string is
> very much about our "empty and NULL are the same string" model.
>
Feels cleaner, and if its cousins behaves that way, copy_string() should
do it to.
> So maybe a simpler patch instead? Something (totally untested!) like:
>
> static inline char *copy_string(const char *s)
> {
> - return s ? strdup(s) : NULL;
> + return (s && *s) ? strdup(s) : NULL;
> }
>
> Hmm?
>
I must have entered vacation mode too quickly and turned off my pointer
fu. It took me 10s to figure out how that one worked.
But, yep, it looks better.
//Anton
--
Anton Lundin +46702-161604
_______________________________________________
subsurface mailing list
[email protected]
http://lists.subsurface-divelog.org/cgi-bin/mailman/listinfo/subsurface
