home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!news.gtech.com!noc.near.net!news.centerline.com!matt
- From: matt@centerline.com (Matt Landau)
- Newsgroups: comp.windows.open-look
- Subject: Re: SunSoft Windows Toolkits Positioning
- Date: 19 Nov 1992 18:27:14 GMT
- Organization: CenterLine Software, Inc.
- Lines: 59
- Distribution: comp
- Message-ID: <1egma2INNsp0@armory.centerline.com>
- References: <l2lco1v@openlook.Unify.Com> <PREECE.92Nov19092817@predator.urbana.mcd.mot.com>
- NNTP-Posting-Host: 140.239.1.32
-
- In <PREECE.92Nov19092817@predator.urbana.mcd.mot.com> preece@urbana.mcd.mot.com (Scott E. Preece) writes:
- >We have had technical contributions from people associated
- >with XVT, Galaxy, both OIs, Wndex, Wcl, THINGS, TAE+, and others;
- >a subset of the XVT API is our base document, but several of the other
- >toolkit APIs are reference documents and we are trying very hard to be
- >inclusive.
-
- Does this imply that the "standard" is going to include several different,
- and mutually incompatible, API's among its base and reference documents?
- What's the point?
-
- >Ironically, with respect to this discussion, we are working on a
- >language-independent API specified as objects and attributes, very much
- >in the style of XView (though we have cut our objects somewhat
- >differently).
-
- Sounds a lot like some components of Project DOE as well.
-
- >standards to be important. The government, of course, also supports
- >open systems and NIST has issued an Application Portability Profile that
- >suggests using a platform independent toolkit for portability. A NIST
- >staffer is a regular participant in the P1201.1 work group, as are
- >several people sent by very large corporations who support open systems.
-
- Yes, but a "standard" that includes three or more different API's seems
- to serve no purpose whatsover beyond letting some vendor certify to the
- government that its toolkit or application "conforms to the standard";
- the fact that such a standard is largely meaningless outside its value
- in the bureaucratic realm of government purchasing paperwork seems not
- to concern the standards bodies at all.
-
- It's like putting two lawyers in a room together -- sure, they'll make
- work for each other, but is that really a worthy goal in itself?
-
- >that the Windows/Macintosh model of the application with local display
- >capability only is the long-term winner. Xt may not be the answer (I
- >personally find Xt a peculiarly unattractive way of expressing
- >programs), but APIs that require a kernel-based, local-display
- >implementation aren't, either.
-
- API's should have NOTHING to say about whether display implementations
- are local and kernel-based, remote and server-based, or some combination
- of the two. A good API should provide a useful, portable way to talk to
- different display systems and input systems.
-
- We can argue about what constitutes "portable" and "useful", and indeed
- that's one of the things standards bodies should be doing. Personally,
- I find the X11 imaging model to be neither, which concerns me even more
- than the rather widely acknowledged shortcomings of the Xt Intrinsics
- architecture.
-
- In any case, an API should specify how you talk to the display and input
- systems -- the details of how the display actually works, and where the
- input actually comes from [the keyboard on your desk, the mouse on the
- other side of the continent, or the laser-guided pointer device on the
- surface of the moon] shouldn't have to affect the programming interface.
- --
- Matt Landau Waiting for a flash of enlightenment
- matt@centerline.com in all this blood and thunder
-