[its-hackers] File name wild cards
eric at swenson.org
Tue Jun 1 19:50:52 CEST 2021
For what it's worth, MIT-MC used a slightly different convention. It, too, used the developer's initials and a version number, but no prefix to indicate the type of file. This provided 2, 3, or 4 digits of version and 2, 3, or 4 characters for initials. This convention was used for Lisp library and Macsyma files. Normally, the directory in which the files were stored provided information as to the kind of file (hence no need for a prefix letter to indicate this). For example, here is a fragment of the files in the LIBDOC directory (from MC):
1 DEBUG* RCW1 5 3/23/2018 17:26:00
1 DEFSTA GJC1 1 3/23/2018 17:26:00
1 DEFVST 164 3 3/23/2018 17:26:00
L DEFVST DOC .INFO. LISP DEFVST
0 DEFVST NIL 3 12/8/1981 16:21:25
1 DIDL CHART 1 3/26/2018 12:08:33
1 DIDL DCH259 17 3/26/2018 12:08:33
1 DIDL HELP 1 3/26/2018 12:08:33
1 DIDL INFO 2 3/26/2018 12:08:33
1 DIRSIZ GSB2 1 3/26/2018 12:08:33
1 DOCTOR JONL2 6 3/23/2018 17:26:00
1 DOW JONL4 1 3/23/2018 17:26:00
All of these were lisp sources.
On 6/1/21, 10:18 AM, "its-hackers on behalf of Jack Haverty" <its-hackers-bounces at its.victor.se on behalf of jack at 3kitty.org> wrote:
Well.... Convention 2 is long gone, so there are no rules anymore. You
can do whatever you want with your file names.
If you're trying to be historically accurate, the typical pattern for
FN2 was a letter often signifying use of file, followed by your login
initials, followed by a 2-digit version number. So, for example, a file
containing PDP-10 assembler might have been "STRPRS MJFH03" for a
3rd-generation version of STRing PaRSer that I (JFH) wrote.
That naming convention mostly applied when you put your files into a
"public" directory for others to use. Of course there was no file
system protection or permissions, so it was just a Convention to do so.
Two digits might not be enough for versions, so I remember keeping work
in progress in my own directory, until it was finished, debugged, and
ready for release - so STRPRS might have started life as DSK:JFH;STRPRS
1 and evolved from there to eventually be "released" as STRPRS MJFH01.
I don't remember all of the prefix letters. M indicated Midas, U for
mUddle. Probably there were some more. I don't think there was a Teco
prefix - a file you were editting was likely an M or U instead.
If what you're trying to do is figure out how to search for and find
files in some directory, the classic way to do that was to read the
.FILE. (DIR) "file" from that directory, and simply parse the results to
get all of the file names. E.G., reading DSK:JFH;.FILE. (DIR) would
get you a listing of all files in DSK:JFH; -- somewhat like doing "ls
-ls" in Unix.
On 5/31/21 4:21 PM, rrs0 at earthlink.net wrote:
> Thanks Jack,
> I may have to change the way I have been creating my FN2's (2nd file name). On Twinx xxx.mud or xxx.nbin or
> xxx.fbin would work. On ITS, I always used URSnnn for my FN2, "U" stood for MDL source code and "RS" are my
> first and last initials. I forgot what the FN2 prefix characters for Midas ("M" maybe) Calico , and TECO were.
> Should I drop the prefix letter and <STRING <SNAME> <UNPARSE .N> where .N is the
> sequence number MOD ( 6 - <LENGTH <SNAME>>)? For me URSnnn would work fine for FN2 and ">".
> Or for the FN2 should I go with " <SNAME>nnn"or "U"<SNAME>nn. I could build in "URSnnn" as special case for "RRS;"
> But that that not a tasteful solution.
> RRS at MIT-DMCG (DMS)
>> On May 31, 2021, at 1:52 PM, Jack Haverty <jack at 3kitty.org> wrote:
>> IIRC, there was never any "wild card" mechanism in ITS, other than the use of <> characters. Most common was just a satiating numeric second name to represent a version number. But for things where you didn't want any ambiguity, some unique FN2 would be used.
>> In the MIT-DM group, there was at one point in the 70s a scheme for documentation established, called "Convention 2" (I don't remember there being a Convention 1). I think that's where the notion was created of second names combining the username with a version number. There was a big push to create "libraries" of software, putting the relevant files into a common directory; prior to that, everyone mostly kept their files in their own directory.
>> But I don't think there wasn't anything special in ITS to treat such names differently. If you used an FN2 of <loginname><version>, the > and < would typically pick out the file you expected. Any specific file was typically created and evolved by one user, so the <loginname> part of that file's name wouldn't change.
>> For what cases does <> not work? I think of <> as like the Unix shell * convention, but selecting the file name with the numeric biggest or smallest result if there's more than one match. So it should always find something if there's any match on FN1. (<> is something I've always wished Unix had too.....)
>> /Jack Haverty
>> JFH at MIT-DM 1970-1977
>> On 5/31/21 11:19 AM, rrs0 at earthlink.net wrote:
>>> Hi Everyone,
>>> Its been 40+ years since MIT-DMS, I have forgotten wild cards in ITS file names. Still remember "<" ">" for the second file name. I am starting to add file system support too "TT:RRS;ITSPKG >", the muddle package system I'm building. I want to
>>> Find muddle packages in my and other directories to load. Picking second file names to search for is the hard part. For me FN1 MUD or FN1 URSxxx wild work. Most of the time FN2 equal to ">" would work, What should I pick for a general
>>> FN2 to search for in the user directories, So far wide don't have the muddle library system to work with.
More information about the its-hackers