jaffe@caip.RUTGERS.EDU (Saul) (07/09/86)
We at Rutgers are currently in the process of eliminating one of our DEC-20 machines. The problem is that we now have to convert many of our users' APL workspaces to another machine. We are moving to a Pyramid (running 4.2 BSD) and plan to use DYALOG APL. Does anyone out there have any experience in transferring workspaces to other machines and APL versions? I am particularly interested in any APL programs that you may have that allow you to move workspaces from one implementation to another. Both the DEC and DYALOG versions of APL have workspace transfer programs but they are intended for use for transferring from one DEC machine to another (or one DYALOG APL to another). Neither program works on the other machine unfortunately. Can anyone help? I would greatly appreciate it. Saul Jaffe Rutgers University ARPA: Jaffe@Rutgers UUCP: caip!jaffe
ljdickey@water.UUCP (Lee Dickey) (07/15/86)
> [we] are moving to a Pyramid (running 4.2 BSD) and plan to use DYALOG APL. Lucky you! Dyalog has the best UNIX-APL interface that I have seen. > Does anyone out there have any experience in transferring workspaces. Yes. The only recommended method is the WorkSpace Interchange Standard. I think that I have done the largest ws conversion ever, about 2200 workspaces, and could not have done it without WSIS. Some ad hoc methods can come close, but there is *nothing* that beats it. Dyalog might have a WSIS receiver, because surely they have faced this problem before. (A number of people are moving to Dyalog.) At least two progressive APL's include WSIS as a built in utility, and maybe Dyalog is one of them, I can not remember. If they do not have it yet, this is a good time to negotiate for it! It is not hard to write a WSIS sender. If you also have to write your own receiver, you can simplify the WSIS conventions because you know exactly what the atomic vector is on each machine. There are a couple of places to look for WSIS documentation. If I remember correctly, it is the Dec 79 issue of APL Quote Quad that has WSIS0. Appendix B of the Draft International Standard has WSIS1.
libes@nbs-amrf.UUCP (07/16/86)
I wrote a WSIS0 implementation when I worked for CCIS at Rutgers. It's probably lying in a vault somewhere. In any case, I wrote a document that included all the source to it. Very little of it is system-dependent - just things like the system commands to access the tape drives, and building the WSIS representation of the atomic vector. Otherwise, you should be able to use it. When I left Rutgers, I dumped some IBM workspaces and was able to read them on a DEC-10 at Rochester that also had a WSIS. I assumed it was DEC's implementation. But from your message it sounds not. Try contacting someone at the UofR comp center. I can't help you with Dyalog. Lee, I never heard of WSIS1. Could you briefly describe what the differences between WSIS1 and WSIS0 are? Don ")314159" Libes {seismo,umcp-cs}!nbs-amrf!libes
ljdickey@water.UUCP (Lee Dickey) (07/22/86)
> Lee, I never heard of WSIS1. Could you briefly describe what the > differences between WSIS1 and WSIS0 are? There are several differences: (1) There is one more class of canonical representation vector: WSIS0 WSIS1 Meaning C C Character array N N Numeric array F F Function P P Pseudovariable Q Executable expression X X Installation dependent The "Q" class is for executable expressions. For example you could have 19Q#fx 2 2 4 FOO A+23 (The "#" stands for an APL "quad".) This class Q could replace C, N, and F, if you wanted it to, but it has the complication of doubling quotes in character strings. Not quite as clean, but it is implemented correctly on some systems. (2) The first Pseudovariable is changed: From "9PWSIS 0 0" to "9PWSIS 0 1". (3) There are two new Pseudovariables that allow representation of an enormous character set. At present, still only the APL char set (regn no. 68) can be used in a conforming program. But DATA may be chosen from any registered character set. (There are many.) (a) BITS this tells how many bits are being used for each character (at least 8). (b) ESCAPE this identifies what character set each element of the atomic vector is taken from. ------------------------------------------------------------------------ I hope that helps. There are more details, of course, but that is a rough outline. You can find out more by looking at APL Quote Quad, Volume 14, Number 2, December 1983. This was the issue that was dedicated to the APL Standard. In it there is something called Annex B, which is devoted to WSIS1. There have been some corrections to the standard since then, but the outline of the standard today is pretty much as it was then. I think that Annex B has not changed, except that it has been re-typeset, and possibly typos may have crept in. But, to get the best copy, you should get in touch with the International Standards Organization, and ask for Draft Proposal 8485, February 25, 1986. As an easier alternative, contact your local national standards committee for the latest version. These are the countries that I know of that have active working groups. Country Standards Group USA ANSI X3J10 UK BSI APL Working Group Canada CSA APL WG France AFNOR APL Group de Travail W.Germany DIN Sweden SIS Belgium ?