Re: [gtk-list] Re: Mac?

[Valdis Kletnieks vt edu]

On Wed, 08 Sep 1999 06:25:03 EDT, John C Atkeson <jcatkeson@juno.com>  said:
>   A Mac port of gtk would instantly expand the customer base of every
> developer on this list.  

No, it only expands the customer base of every developer who writes
code that is portable *except for the GTK part*.

Unless you have a *really* good POSIX emulation package for your Mac,
I suspect you're doomed.  Hint - what does the 'group' and 'other'
fields mean in the st_mode field returned from a stat()?  What do you
do with code that assumes the bits mean anything?

I'm sure equally evil things are waiting for any programmers of
things like   printf("Username is %s",getpwuid(getuid())->pw_name).

Question 1:  Do you have a getuid() and a getpwuid() that do anything
reasonable?

Question 2: Does the code make unwarranted assumptions regarding
the output of question 1? (Note, this is a *lot* trickier than you
think - I've seen *unix* programs that lose on the getpwuid/getuid
stuff, if run set-UID, or if there's more than one entry in /etc/passwd
matching that UID, etc etc etc.

Just having the toolkit does NOT make everything warm and fuzzy. It's
only a start.

Also, there *will* be tradeoffs.  I suggest that anybody who is
seriously looking at a Mac or Windows port go look at the Tk/Tcl
stuff, or the Perl source code, to see the sort of issues that you
will run up against.  Tk in particular had to make a lot of sacrifices
in order to be cross-platform portable.

				Valdis Kletnieks
				Computer Systems Senior Engineer
				Virginia Tech