phil@ux1.cso.uiuc.edu (06/07/89)
A UNIX version of REXX would make sense. It may not be everyone's favorite language, but it does have some powerful capabilities. One would only need to write (in C) a REXX interpreter that took as its first argument the name of the REXX file to run. Then the REXX syntax would be modified slightly to accept #!/{path}/rexx at the beginning of the source file. You could then make "rexx scripts" just like shell scripts, only in a different language. Further extensions to give access to UNIX functions would also be nice (remove all the CMS-isms). Does some wizard of writer of language interpreter want to tackle this one? --Phil howard-- <phil@ux1.cso.uiuc.edu>
terry@uts.amdahl.com (Lewis T. Flynn) (06/07/89)
In article <7800001@ux1.cso.uiuc.edu> phil@ux1.cso.uiuc.edu writes: > >A UNIX version of REXX would make sense. It may not be everyone's favorite >language, but it does have some powerful capabilities. [suggestions deleted] >Does some wizard of writer of language interpreter want to tackle this one? > >--Phil howard-- <phil@ux1.cso.uiuc.edu> I've been trying to convince Charles Daney (the author of Personal REXX from Mansfield Software) to do just that. So far, he's unconvinced that there is a big enough market to do a UNIX port. Any ideas of how many he might sell at a reasonable price (the ms-dos version lists for $125)? Terry Flynn #include <std/disclaimer>
brooking@mcnc.org (James A. Brooking) (06/09/89)
In article <7800001@ux1.cso.uiuc.edu>, phil@ux1.cso.uiuc.edu writes: > > A UNIX version of REXX would make sense. It may not be everyone's favorite > language, but it does have some powerful capabilities. > ... > --Phil howard-- <phil@ux1.cso.uiuc.edu> I manage a Cray installation (X-MP/28) which runs UNICOS, the Cray unix variant. We also support a number of unix- based workstations (SUN, SGI) and a few dozen VAXen of various flavors. Previously I worked at a small, family-owned business (GE) as a sort of chief VM guru, where I first encountered REXX. I'm here to tell you that REXX is an invaluable tool for "front-ending" code with poor user interfaces (although we all know there isn't much of THAT around...) and replacing ill-conceived command names or calling arguments by something fit for human consumption. It also very useful, as other articles have mentioned, for prototyping. And in fact, at least in a VM environment where the REXX interpreter is quite efficient, the prototype can become the production version with a clear conscience. Others may have mentioned as well, that REXX (at least in the VM and MS-DOS (Mansfield Software versions) has an excellent debugging facility. This, in combination with the language's inherent ease of learning and use, makes for a development tool that MAKES ITS USERS LOOK GOOD! REXX programmers (at least this one) can generate debugged code at an unheard-of rate. My favorite story is about a task assigned to a "C" programmer and me to create a generic batch facility by developing cooperative processes on the 3081 (me, REXX) and a SUN workstation (him, "C"). His client software took a couple of months to "complete" (it was still kind of buggy, in the style of unix). Mine took a week, and hasn't been touched in the three years since it was written. The point is that a REXX capability for unix (preferably portable to UNICOS) and for VAX/VMS would be a very welcome capability in my shop... I'd pay well for it. 'Nuff said. Jim Brooking Technology Applications, Inc. (401)841-5354 brooking@nusc.navy.mil
zjat02@apctrc.trc.amoco.com (Jon A. Tankersley) (06/10/89)
I think that Mansfield is missing something here. Rexx for UNIX is going to be coming. The IBM commitment to UNIX, if it is sincere, will almost require it for a familiarity aid/conversion tool for VM hackers. It is rumored that there are some Rexx compilers in the works somewhere. I imagine we just have to be patient. It might also be nice to have Xedit on UNIX. It is easier to learn than vi and would be another familiarity/conversion aid for IBMers that see the light :-). -tank- #include <std/disclaimer.h> /* nobody knows the trouble I .... */ tank@apctrc.trc.amoco.com ..!uunet!apctrc!tank
clewis@eci386.uucp (Chris Lewis) (06/13/89)
In article <4678@alvin.mcnc.org> brooking@mcnc.org (James A. Brooking) writes: >... And in fact, at least >in a VM environment where the REXX interpreter is quite >efficient, the prototype can become the production version >with a clear conscience. As another testimonial, I implemented *all*[1] of UNIX System V "make" (greatly extended from V7) in REX (on VM/SP) in 2-3 days in about 1000 lines of code. Ran quite fast. Used by a number of people on the IBM network. Was more than slightly surprised at the power of REX and its speed. [1] With the exception of archive handling (I actually did implement it, then discovered to my horror that TXTLIB's don't have dates) and SCCS support. PS: don't ask me for it - I don't work at IBM any longer and don't have the program anymore. Sorry... -- Chris Lewis, R.H. Lathwell & Associates: Elegant Communications Inc. UUCP: {uunet!mnetor, utcsri!utzoo}!lsuc!eci386!clewis Phone: (416)-595-5425