kurt@PRC.Unisys.COM (Kurt Wallnau) (03/16/90)
Greetings, I'm chief programmer of the Unisys STARS task which is developing an Ada implementation of the MIT X Toolkit. We've recently received a spate of requests for information on our Ada/Xt project, such as the request below. At this point in our project it is probably fair to talk about what we've done, what we'll have, and when it will be available. >> From: Tucker Taft <taft@ajpo.sei.cmu.edu> >> Teri Payton at Unisys (she is their chief STARS architect) >> is directing the development of an implementation of Xlib and Xt >> in Ada. It looks pretty good. >> Her E-mail address is payton@stars.reston.unisys.com. > From valdi@kerfi.hi.is Wed Mar 14 04:41:36 1990 > To: payton@stars.reston.unisys.com > Could you comment on the status of this project e.g. > 1) Estimated date of project completion ? > 2) Supported packages (librarys) e.g. Xlib, Xt, Motif etc. ? > 3) Software will be public, shareware or sold ? > 4) Cooperation with MIT ? First, it might be useful to describe what we are (and aren't) doing with X in Ada in this "increment." Our task is to develop an implementation of the MIT X Toolkit, in Ada. Our basis is the X11R3 Xt. An upgrade to the X11R4+ toolkit has been discussed, but as yet there are no concrete plans. Although an implementation, in Ada, of xlib would be very desirable, for practical considerations we decided to "make do" with the Ada bindings to xlib done by SAIC under a STARS Foundations contract. We've extended the bindings to include some missing interfaces, and made some other minor syntactic changes. Within STARS we are currently grappling with the issue of Ada/Xlib standardization - perhaps this, too, is a valid net discussion topic. Our approach to developing an Ada/Xt implementation has been constrained by two (sometimes competing) objectives: 1) To develop an implementation which is manifestly equivalent, functionally, with the MIT X Toolkit. In practice, this means providing all of the features of C/Xt in such a way that there is a clear, intuitive mapping from Ada/Xt to C/Xt. 2) Make appropriate use of Ada language features to develop a portable, compiler-independent toolkit, and to provide a programming interface which is representative of good Ada style. Ok, the above two points sound like "motherhood." The important point is to note that 1 and 2 sometimes appear to be in conflict. We've already written a fairly extensive design report on the subject of our Ada/Xt work, and so I am hesitant to repeat gory details here, which in any event may not be of general interest. I will state, however, that I think we have developed a very good mechanism for widget subclassing, and a reasonably good "simulation" of procedure types as abstract data types. I can certainly discuss these points in more detail if there is interest. The most important conclusion is that judicious use of Ada LRM chapter 13 features and information hiding can provide very reasonable solutions to problems imposed by the "C" bias of the Xt model. And now, the specifics of the query: >Could you comment on the status of this project e.g. >1) Estimated date of project completion ? We are scheduled to deliver code on April 12, May 12, and June 12. The April release is intended to demonstrate intrinsics functionality on a (very!) small set of sample widgets. The May release is intended to provide a more fully fleshed-out widget set, in particular as driven by another STARS project which is developing a graphical browser of ERA-style information models. The final release will be what we will consider a suitable basis for inter-STARS standardization. I am hopeful that industry at large will consider our work a reasonable basis for beginning standardization. >2) Supported packages (librarys) e.g. Xlib, Xt, Motif etc. ? We will be providing Xlib bindings, the X11R3 intrinsics (specs and bodies), pre-defined widget classes (public/private specs, bodies), and a sample widget set. The exact widgets to be developed have not been fixed, but will be derived from the Athena, HP, and Sony widgets which appear on the R3 release tape. We will not be supporting Motif - this would require extensions to the intrinsics and some uncertain licensing issues. Of course, it may be that R4+ will converge at the intrinsics level with OSF... >3) Software will be public, shareware or sold ? Beats me. I think it is publically available but there may be export controls. You should contact Teri Payton, the Unisys STARS Systems Architect, for details (payton@stars.reston.unisys.com). >4) Cooperation with MIT ? We presented our plans to the consortium last April. We noted a lukewarm but polite response to the idea of developing an Ada/Xt implementation. Our ultimate goal is to have the consortium distribute (if not maintain!) our implementation. We shall see. > >| Thorvaldur Egill Sigurdsson | Internet: valdi@kerfi.hi.is | >| Engineering Research Institute | | >| University of ICELAND | Phone: 354-1-694699 | >| Hjardarhagi 2-6 | | >| 107 Reykjavik, ICELAND | | >------------------------------------------------------------------------ Kurt Wallnau Unisys, VFO 70 E. Swedesford Rd. Paoli, Pa. 19301 (215) 648 2713 kurt@prc.unisys.com