Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!seismo!rutgers!ucla-cs!zen!ucbvax!DECWRL.DEC.COM!haynes From: haynes@DECWRL.DEC.COM Newsgroups: comp.windows.x Subject: Re: Fixes to paint and xedit in V11.2 Message-ID: <8707280329.AA01894@vulcan.dec.com> Date: Tue, 28-Jul-87 00:20:01 EDT Article-I.D.: vulcan.8707280329.AA01894 Posted: Tue Jul 28 00:20:01 1987 Date-Received: Wed, 29-Jul-87 02:25:00 EDT Sender: daemon@ucbvax.BERKELEY.EDU Distribution: world Organization: The ARPA Internet Lines: 31 Being one of the primary designers and implementors of the "Xt" toolkit, I'd like to respond to the recent rash of criticisms of null pointer handling in the toolkit. First, please do recognize the distinction between the applications (clients) and the toolkit itself. The bugs that Jordan and Ed reported were both in clients (xedit and paint). I can't control what other people do with our code. Second, I don't know of any places in the toolkit that allow passing NULL to the string routines. If any such exist they are bugs. PLEASE REPORT THEM!!! We can't fix bugs that we never see! I have see NO that is NOT ONE bug report from the people complaining about null handling in the toolkit. You want to gripe, or you want to be helpful? Send your bug reports to Todd at athena, and they'll get sent to the right people. By the way, passing a null string as a command label is illegal. Programs that do this should die (and will on a 68k). You can argue whether or not this is good, but that's the intended behavior. So the fact that we blithely pass such strings to strcmp etc. is not evidence of our VAX/Unix chauvinism, it's evidence of our laxness in documentation. Which flaw I freely admit and unashamedly ask for help with! Any of you kind folks out there willing to DONATE time to help us write toolkit documentation? After all, I'm personally making big bucks sitting here answering mail about the toolkit... "If you can't make 'em fix it..." grrr... -- Charles