exspes@gdr.bath.ac.uk (P E Smee) (05/14/91)
Well, one thing this 'what happened to Multics' discussion has brought out is that there are a lot of ex-Multicians hiding out there in the Unix woodwork. So, has anyone out there got a PD or personally-implemented substitute for the Multics 'memo' command, running on Unix, which they'd be willing to share? Misc notes: 'calendar' is 'close, but no cigar'; not nearly powerful enough. Though it might be possible to script up a memo-substitute from it and some assorted awk-wardness. (If you can't do something with a line of awk, it's probably not computable anyway. :-) In the interests of coexistence with Unix, I could live without alarm and call memos if required (though they would be nice). I want -date, -expires, -retain/noretain, -repeat (and -repeat_when_processed), and -invisible; and -print/-list of -all or mature only. (All as functionalities. It's a matter of some indifference whether adding, deleting, and modifying memos is done via the hypothetical command/subsystem, or by vi-ing the memo-file analog.) Oh, and I want to be able to include dates in the memo text which will not activate the memo. E.g. a memo to be printed from May 21 to May 24, repeat each year, text saying 'Annie's birthday May 25'. -- Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132
exspes@gdr.bath.ac.uk (P E Smee) (05/18/91)
In article <1991May14.161704.27609@gdr.bath.ac.uk> P.Smee@bristol.ac.uk (Paul Smee) writes: >Well, one thing this 'what happened to Multics' discussion has brought >out is that there are a lot of ex-Multicians hiding out there in the >Unix woodwork. > >So, has anyone out there got a PD or personally-implemented substitute >for the Multics 'memo' command, running on Unix, which they'd be >willing to share? Well, so far in response to this I've had a (not very encouraging) pointer to a thing called 'calend' which I might be able to pick up from one of the archive sites. If that was all, I wouldn't bother replying to myself, but I've received a number of pieces of mail which fall into two categories which might yield to outside help. (If I knew which newsgroup the posters had seen this in, I'd limit my reply, but since I dont...) First, a number of people have asked me if information about the Multics command set (and in one extreme case, also the subroutine library) is available. Preferably online. I don't know, does anyone? (I kept my copies of the commands manual, and of the subroutines library, but as the first is about 1100 pages long, and the second even bigger, I'm not ABOUT to type them in. :-) And, I'm no longer in a position to resurrect the info files from the distribution tapes. Anyone got this stuff online? Will HIS still sell manuals for their non-existent implementation of computer-nirvana? Second, I've had 3 people reply along lines of 'that [memo] sounds like it would be fun to try writing, can you give me a description of how it works?' Since I'm always happy to help people out where it requires little effort on my part (and particularly when the results might satisfy one of my perceived needs) a bulk answer: I'll generate a reasonable description of its black-box behavior (don't want to pre-empt implementation details) and mail it to you guys (and/or post it, if I get any more such requests over the weekend). Do note it's non-trivial, primarily because the standard Unix date/time handling library routines aren't up to the task, so you'll have to invent something along those lines as well. Cheers... -- Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132
gintell@granite.ma30.bull.com (John Gintell) (05/20/91)
There is a company called ACTC in Calgary, Canada - a spinoff from the University of Calgary - that built some UNIX software to help convert Multics users into UNIX users. They built a considerable amount of software, quite likely including memo. I don't know what its status is and whether they would give it away or sell it, but you might try contacting them. The President of ACTC is Arun Gatha; the chief technology person is Tom Oke. They can be reached at 403-295-5900. You should be able to reach Tom Oke via email as oke@system-m.az05.bull.com ..yes, system-m is Multics .... _______________________________________________________________ John W Gintell Groupe Software Technology Bull HN MS 821A 300 Concord Road Billerica, MA, 01821 (508)-294-4181 gintell@granite.ma30.bull.com
dan@cpsc.ucalgary.ca (Dan Freedman) (05/26/91)
In article <1991May20.151334.2369@granite.ma30.bull.com> gintell@granite.ma30.bull.com (John Gintell) writes: >There is a company called ACTC in Calgary, Canada - a spinoff from the >University of Calgary - that built some UNIX software to help convert >Multics users into UNIX users. They built a considerable amount of >software, quite likely including memo. I don't know what its status >is and whether they would give it away or sell it, but you might try >contacting them. As a long-time Multics fan, and as one who still has an account on the University of Calgary's Multics system, I would be happy to try to answer people's questions about Multics. The ones I can't answer myself I can pass on to the people at ACTC. Dan Freedman -- U. of Calgary Computer Science Dept., 403 220-7299 2500 University Dr. N.W., dan@cpsc.ucalgary.ca Calgary, Alberta, Canada. T2N 1N4 VE6DFM
fetter@cos.com (Bob Fetter) (05/29/91)
In article <1991May17.173657.29391@gdr.bath.ac.uk> P.Smee@bristol.ac.uk (Paul Smee) writes: > Do note it's >non-trivial, primarily because the standard Unix date/time handling >library routines aren't up to the task, so you'll have to invent >something along those lines as well. Actually, a Unix implementation of convert_date_to_binary_ would IMHO be the most useful product of this type of effort. I would have to think, though, that someone, somewhere, has done this already. The usefulness of dealing with date/time strings like next friday at noon two weeks after easter tomorrow at 9pm would be nice. An "extention" to the Multics notation, though, would be negative times -- such as "last Thursday". I don't think (remember) the Multics routines handling negative offsets. Did it? The last I looked at the Multics implementation, it was in rdc (reduction_compiler) format (akin to 'yacc' on Unix, but done before yacc came about <I think>). Ah, this brings up a tangent... The "Yet Another Compiler Compiler" indicates that it's an offspring of earlier work. Does anyone know what the earlier work/compiler_compilers were? -Bob-
exspes@gdr.bath.ac.uk (P E Smee) (05/30/91)
In article <45901@cos.com> fetter@cos.UUCP (Bob Fetter) writes: > > Actually, a Unix implementation of convert_date_to_binary_ would IMHO >be the most useful product of this type of effort. I would have to >think, though, that someone, somewhere, has done this already. > > The usefulness of dealing with date/time strings like > > next friday at noon > two weeks after easter > tomorrow at 9pm > >would be nice. An "extention" to the Multics notation, though, would >be negative times -- such as "last Thursday". I don't think (remember) >the Multics routines handling negative offsets. Did it? Yep. In later incarnations it also knew about 'before' and 'after', so (to pick an example from the manual) Tuesday after Monday on or after 11/1 is Election Day in the US -- i.e. the Tuesday after the first Monday in November. The baroque Tuesday after Monday on or after 11/1 -1 week would be the Tuesday before that, as would Tuesday before Monday on or after 11/1 'Easter -2 days' is Good Friday. -- Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132
forrie@morwyn.UUCP (Forrie Aldrich) (05/31/91)
I have seen this discussion for some time now, and am really curious if
there is a pd command implementation that does this 'memo' command out
there anywhere... I think that was the initial question, no? If there's
something out there, please leave the FTP address or UUCP address to
get it... thanks.
Forrie
--
--------------------=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--------------------
Forrest Aldrich, Jr.| (a reliable path here someday) |forrie@morwyn.UUCP
| <email paths> |
CREATIVE CONNECTIONS| uunet!virgin!unhtel!morwyn!forrie |Graphic Illustration
------------------\-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=/------------------
\___ PO Box 1541 - Dover, NH 03820 ___/ fetter@cos.com (Bob Fetter) (06/01/91)
In article <1991May30.103104.24322@gdr.bath.ac.uk> P.Smee@bristol.ac.uk (Paul Smee) writes: >Yep. In later incarnations it also knew about 'before' and 'after', so >(to pick an example from the manual) > Tuesday after Monday on or after 11/1 >is Election Day in the US -- i.e. the Tuesday after the first Monday >in November. The baroque > Tuesday after Monday on or after 11/1 -1 week >would be the Tuesday before that, as would > Tuesday before Monday on or after 11/1 > >'Easter -2 days' is Good Friday. Is there (E)BNF available for the date/time strings that can be shared outside of Bull? With that, munging the details (like easter) should be straightforward/at least do-able. Also, isn't there a convert_date_to_binary_$relative_time rtn (or something similar) which just parsed time deltas from current time? I'm thinking of memo's "-repeat 2 weeks" argument. Or was that just a degenerate case of not giving a base time along with the delta in the string? It *has* been a while (longer than I really care to think about)... -Bob-
exspes@gdr.bath.ac.uk (P E Smee) (06/03/91)
In article <45919@cos.com> fetter@cos.UUCP (Bob Fetter) writes: > Also, isn't there a convert_date_to_binary_$relative_time rtn (or >something similar) which just parsed time deltas from current time? >I'm thinking of memo's "-repeat 2 weeks" argument. Or was that just >a degenerate case of not giving a base time along with the delta in >the string? It *has* been a while (longer than I really care to think >about)... Almost, but backwards. cdtb_ took a character string and gave back a fixed bin (71) system date-time and an error code. cdtb_$relative took a char string and an fb(71) system date-time, and gave back an fb(71) system date-time and an error code. Both worked the same if the character string was an absolute date/time, e.g "10:15 AM March 3 1992". If the string was a relative date/time (e.g. "March 12", which meant the NEXT occurance of March 12 after the reference date/time, or "+5 days" which means 5 days from the reference date/time) then cdtb_ would do the calculation using the current system clock as the reference time, while cdtb_$relative would use the passed-in clock value as the reference date/time. In fact, as I recall, since PL/1 allowed multiple entry-points to a procedure, the implementation was simply that cdtb_ read the system clock and then fell through into the cdtb_$relative case. The epoch was set at 0000. GMT 1 Jan 0001 AD, and cdtb knew about dates from Jan 1 0001 AD thru to Dec 31 9999 AD, resolution down to the microsecond. (It is pointed out as a limitation, though, that it did NOT know about leap-seconds.) -- Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132
cwitty@cs.Stanford.EDU (Carl Witty) (06/04/91)
In <1991May30.103104.24322@gdr.bath.ac.uk> exspes@gdr.bath.ac.uk (P E Smee) writes: >In article <45901@cos.com> fetter@cos.UUCP (Bob Fetter) writes: >> Actually, a Unix implementation of convert_date_to_binary_ would IMHO >>be the most useful product of this type of effort. I would have to >>think, though, that someone, somewhere, has done this already. >> >> The usefulness of dealing with date/time strings like >> >> next friday at noon >> two weeks after easter >> tomorrow at 9pm >> >>would be nice. An "extention" to the Multics notation, though, would >>be negative times -- such as "last Thursday". I don't think (remember) >>the Multics routines handling negative offsets. Did it? The date parser included with the Andrew Message System, in .../ams/libs/ms/prsdate.y, handles all of the above dates (including "last Thursday"), except that it doesn't know about Easter--I tried "two weeks after christmas" instead. >Yep. In later incarnations it also knew about 'before' and 'after', so >(to pick an example from the manual) > Tuesday after Monday on or after 11/1 The parser in prsdate.y won't handle this (it doesn't know about "on or after"), but it does handle the equivalent "tuesday after monday after 10/31". If it annoys you to have to remember that October has 31 days, you could also use "tuesday after monday after day before 11/1". >is Election Day in the US -- i.e. the Tuesday after the first Monday >in November. The baroque > Tuesday after Monday on or after 11/1 -1 week That doesn't work. >would be the Tuesday before that, as would > Tuesday before Monday on or after 11/1 That does. (Except that "on or after" doesn't work.) So does "one week before tuesday after monday after day before 11/1". >'Easter -2 days' is Good Friday. >-- >Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK > P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132