jdb@mordor.UUCP (John Bruner) (02/13/84)
I am very unhappy with Berkeley for making a change to "my" APL for the 4.2BSD distribution. In the fall of 1982 I sent a tape of the Purdue/EE version of UNIX APL to Berkeley for possible inclusion in the 4.2BSD distribution. Since that time I asked several times what changes were necessary to prepare it for 4.2; I never received a reply. At one point I was contacted by Berkeley to obtain my current address and a description of APL. Again I asked for information about changes made to APL and again I received no answer. Unfortunately, changes were made, and they are incompatible with previous versions of the software. The format of workspaces has changed subtly, just enough to make old and new workspaces incompatible. The seed for random number generation has been included in the group of miscellaneous variables that are stored at the beginning of a workspace file. One of my goals with APL was to preserve compatibility as much as possible between implementations, although the program is ancient (it predates version 6 UNIX). APL still runs (at least my version does) on PDP-11's with I/D space, and I can convert workspaces between PDP-11 and VAX formats. I don't know if the 4.2BSD distributed version will run on 11's (I'll try to find out), but since its workspace is incompatible with previous workspace formats you are sunk if you want to continue using old workspaces. I'm not arguing that the change was undesirable in itself, but a new magic number should have been assigned to workspaces with the newer format. Had I known about this I would have put in code to read older-format workspaces. Now there is little hope that a clean solution can be worked out. I'd like to ask everyone who has 4.2 to change the magic numbers in "apl.h" to something else, but of course I could never reach everyone. The people who suffer the most will be at Purdue/EE. Since the APL development was all done there, there are lots of old, now-incompatible workspaces. If they adopt the 4.2BSD release all of the old workspaces stop working. If they don't they are incompatible with the rest of the world on a program which (in recent years anyway) was DEVELOPED AT PURDUE/EE. I'll take care of Purdue/EE; though it wasn't my fault I hate to leave them in the lurch. For any others who have one of my distribution tapes of APL I'll try to post a little program that will convert workspace formats. Of course, you can't tell the old from the new since they have the same magic number, but it's the best I can come up with. Finally, I'm sorry to say that I am not in a position to distribute APL. Working on APL was always a spare-time project (it is even more of one for me now). Things are even more complicated at the moment because our site is still running 4.1 I have been told that there are some other flakey problems with APL under 4.2 If bug reports are posted I can't promise fixes but I will look into them. -- John Bruner (S-1 Project, Lawrence Livermore National Laboratory) MILNET: jdb@s1-c UUCP: ...!decvax!decwrl!mordor!jdb