perm filename MSGLOG.MSG[S,AIL]15 blob
sn#098120 filedate 1974-04-21 generic text, type T, neo UTF8
∂11-APR-74 1448 network site MAXC
Date: 11 APR 1974 1452-PDT
From: SPROULL at PARC-MAXC
Subject: A BUG
cc: RS at SU-AI
- - - -
The following bug, ompiled at the AI project with the /H
siwtch on , or compiled here at Xerox just normally (remember,
/H is standard on TENEX SAIL), gives the "Virual origin ---
horseshit message. However, thee is no trouble with the
virtual origin -- the code comiled is fine, etc.
(I remember finding a bug relating to this section of things
when I took SAIL to NIH -- this error message was printed
out when it should not have been. Is is possible that the
edit was lost?? I peered vbriefly at the ARRAY code on S,AIL
and couldn't remember enuf context to decide whether the fix
was there, or lost....)
Here is the code:
00100 begin "classify"
00200 integer uttn,i,j,k,l,m,n,iword,iutt,fnum,flen,lnum,llen,ds,
00300 depth,obr,descthresh,pdtol,dist,iw,ld,br,lt,utts,i1,
00350 iclass,ndesc,fb,nt;
00400 integer array big[1:20860,1:4],cs[1:13820],classes[1:55],
00500 where[1:40],ns[1:100],ibest[1:10];
00550 real res1,res2;
00600 real array result[1:55],iresult[1:55],best[1:10];
00700 string word;
00800 string array thewords[0:100];
00850 end;
So, that's it. Have fun.
Bob Sproull
-------
∂31-MAR-74 2332 THE,JRL
no error message with COP(listitemvar)
∂06-APR-74 1251 1,RS
(1) I DECIDED TO COME UP HERE SO THAT FILES WOULD
BE IN TV FORMAT WHEN YOU GET THEM, THUS CREATING LESS CONFUSION.
(2) I TRANSFERRED TOTAL OVER TO [S,AIL]. PERHAPS
I SHOULD HAVE WAITED BUT IT WAS SO LITTLE DIFFERENT.
(3) SAIL AND GOGOL HAVE LIKEWISE SOME LITTLE PATCHES
IN THEM. THE PATCHES ARE IN THE FILES ON [10X,AIL], I GUESS
I'LL LEAVE THEM FOR YOU TO SRCCOM AND TRANSFER. GOGOL HAS
1 LINE CHANGED [EXCEPT FOR DIRECTORY-VERSION PAGES], AND
SAIL HAS ABOUT 25 LINES CHANGED.
ALL OF THE ABOVE ARE UNDER TENX CONDITIONAL
ASSEMBLY SWITCHES, SO THERE SHOULD BE NO CHANCE FOR IMPACT.
ALSO, EVERYTHING IS IN TV FORMAT.
BOB SMITH
E
∂05-APR-74 0025 ACT,REG
R SAIL
←foo. ←←←←this shouldn't be an error!
SAIL COMMAND ERROR ....
String space exhausted unexpectedly. ←←←← this probably should be fixed
∂27-MAR-74 1943 network site MAXC
Date: 27 MAR 1974 1945-PDT
From: SPROULL at PARC-MAXC
Subject: Sail bugs
- - - -
Here are some more SAIL things:
1. The export world complains about the description of FILEINFO,
on page 43 of the manual. It says it returns a 6-word block,
and the list of things gives 7 items. Furthermore, these are not
correct in the export world (10/50 or TENEX). Suggest a comment
about what happens in wexport world. For example: "FILEINFO
fills the array with the results of the last LOOKUP or ENTER;
consult the 10/50 manual for the format of these entries."
2. The following program (which comes from an example on page
30 of the manual) does not work, i.e. gets a syntax error inside
the second remember....
ε00100 begin
00200 integer number; context context1,context2;
00300 remember number in context1;
00400 remember context1 in context2;
00500 end;
00600
That"s it.
Bob
-------
∂20-MAR-74 0848 1,RS
WELL IT SEEMS TO WORK. THE FOLLOWING FILES EXIST:
(1) SAILRS.REL
(2) SAILRS.DMP
THE .DMP WAS LOADED /Y/VSAILRS, WHICH I THINK GETS RAID, BUT
IT MAY NOT BE THE CURRENT WAY YOU LOAD THESE THINGS.
IT SEEMS TO WORK ON SEVERAL PROGRAMS. I COMPILED AND LOADED AND
TESTED PUB WITH IT, AND THE RESULTING .DOC FILE SRCCOMED
TO ONE DONE WITH SYSTEM PUB. SO IT LOOKS OK.
SEE YOU LATER,
BOB
∂19-MAR-74 1413 ACT,REG
If you're planning on putting up a new SAIL soon, why not make the
change to REQUIRE LIBRARY and REQUIRE LOAD MODULE that we talked about?
∂19-MAR-74 1315 network site MAXC
Date: 19 MAR 1974 1314-PDT
From: SPROULL at PARC-MAXC
Subject: MAGTAPE WRITER
- - - -
THROUGH THE MIRACLES OF MODERN COMMUNICATION, HERE I AM AT UTAH,
LOGGED INTO THEIR SYSTEM, SENDING YOU A MESSAGE.!!!!
I ABANDONED MY SPECIAL MAGTAPE WRITER. INSTEAD, I WRITE TAPES
IN STANDARD TENEX DUMPER FORMAT (THIS IS A FORMAT THAT EVERY
TENEX SITE CAN READ). THEN, I WROTE A SAIL PROGRAM THAT CAN
READ SUCH TAPES ON ANY SYSTEM (INCLUDING, INCDENTALLY, A TENEX
ONE). HOWEVER, THIS IS NOT QUITE WHAT IS NEEDED FOR BOOTSTRAPPING,
BECAUSE THE SITE WILL NOT HAVE SAIL WITH WHICH TO RUN THE
READER POGRAM. IF, HOWEVER, THE ONLY THING NEEDED IS TOGET A
NEW VERSION OF SAIL UP, THEN THIS WORKS FINE.
SO LET ME KNOW.... I WILL BE BACK IN PA BY MONDAY.
BOB
-------
∂16-MAR-74 1337 network site MAXC
Date: 16 MAR 1974 1337-PDT
From: SPROULL at PARC-MAXC
Subject: -1's
- - - -
Yes, Dan and I are the source of the -1's. At the time (in the distant
past) it sounded as if that was the right thing to put there to plese
DEC. Unfortunately, DEChas changed its mind several times since then.
Bob
-------
∂12-MAR-74 1626 network site MAXC
Date: 12 MAR 1974 1622-PDT
From: SPROULL at PARC-MAXC
Subject: Sail bugs
- - - -
Here are two SAIL bugs for you:
1. The compiler does not handle -5 mod 1 correctly (yes, Virginia, both
of those are constants). It does a 'movsi 230007' as the
answer!! (Where it gets that number from, I don't care to
even guess).
2. CVO will not read the most negative number (12 7's) because the
IMUL gets overflow when the sign flips. This
causes an overflow trap (when trigs running), and an error
message. The cure (I think -- I fixed this a NIH years ago,
and I don't quite remember, so you probably want to think about
it a bit) is to put JFCL .+1 under the IMUL, so that the
overflow code leves well enough alone.
Bob Smith's new SAIL is really great here. Boy is it slick.
So nice. Productivity up 3% overnight. Hope the merge goes smoothly
. Let me know if I can help.
Happy Spring!
Bob
-------
∂12-MAR-74 0310 network site MAXC
Date: 12 MAR 1974 311-PDT
From: SWINEHART at PARC-MAXC
Subject: SAIL BUGS
- - - -
Have trivial fix for non-constant null-string bug, very non-trivial
fix for the suspected (now confirmed) flaw in your copy-to-top
fix (which is in general very nice, not really a kludge, and helps
a lot in my ultimate fix -- the macro body formal search problem again,
of course). This time I really have the fix. THe problem has always
been my saving multiple string pointers on the stack so GC
would adjust all together. Now I will just save counters,
and let substring operator regenerate BPs at right time --
will save but one string, and your PNAME fix will do the rest --
I'll try to get that into a copy of SYM manana, perhaps try it out somehow.
Saved one instruction.
anyhow, talk to you tomorrow.
DCS
-------
∂11-MAR-74 1937 CMU,AIL
People,
We have a new IOSER that mostly reflects some CMU features
that are within CMUSW -- it also has a couple of minor error msg
improvements. We would very appreciate if someone could bless this
copy and make it be IOSER on S,AIL, so that we won't have to
edit these changes in in the future. The new version is
IOSER[CMU,AIL]. We have labeled our changes with =b4=, all within
the CMUs.
Please see if this can happen before anymore changes are made
to IOSER at your end.
Would appreciate a confirmation when all is well.
thanx,
LDE
∂12-MAR-74 2110 network site CMU
***** FTP mail from [A610HS11] (HEARSAY-2)
JIM,
DCS CLAIMS HE HAS A FIX TO BOTH EXTANT
STRING GARBAGE COLLECTOR PROBLEMS AND THAT HE WILL GET
THEM IN PRETTY SOON. (HE JUST TOLD ME ABOUT THIS TODAY.)
IT WOULD BE NICE IF WE COULD GET THOSE FIXES %IN BEFORE THE
TENEX MERGER HAPPENS.
THANX,
LEE
∂6-MAR-74 1453 network site CMU
***** FTP mail from [A700SA00] (SAIL)
CMU bug report, with (hopefully) a fix: =B3= [LEPRUN /36] LDE (6-Mar-74)
FP1 (single word free list) getting fouled up. Problem seems
to be a MOVEM FP,FP1(TABL) near ARYL+5 that looks like it
wants to be a HRRM.
I am somewhat leary of all this stuff, so would appreciate a
conformation of this ASAP.
thanx,
Lee
∂23-FEB-74 1258 network site CMU
CMU bug report. =B1= LDE 23-Feb-74.
Seems that itemvar arrays get allocated looking
like string arrays if they are declared as STRING ITEMVAR ARRAY
or STRING ARRAY ITEMVAR ARRAY. See ITBUG.SAI[A700SA00] or [CMU,AIL].
Besides being wasteful of space (twice as long), the accesses to it
are wrong for non-built in ones, etc.