From: DP 2-FEB-1985 00:21 To: GJC Subj: more on put well it doesn't hang now, but it drops the last pkt. (I think I need to tell the lm to send it, it doesn't appear to have sent) btw: putfil doing a call per char is a real pig, it can consume 30% of the cpu. the lambda seems to be doing a lot of re-trying of packets, I found the same packet in the log file 3 times, (with several ack's preceeding it...) jeff From: DP 2-FEB-1985 00:47 To: GJC Subj: put well i sent a file lambda->vax then retrieved it vax->lambda all of it came back, but it came back double spaced.... jeff From: DP 2-FEB-1985 00:48 To: GJC Subj: double space the vax doesn't have it stored as double spaced... mut be ge... j From: LPH 2-FEB-1985 12:48 To: GJC Subj: RHB His timing tests in DOE-MACSYMA show that the current interpreted version on this 750 is faster than the old NIL-MACSYMA on a 780! From: LPH 3-FEB-1985 10:09 To: GJC Subj: talk at symbolics The linkage of the NAG numerical routines to MACSYMA via some LISP NAGLINK will be the subject of a talk at Symbolics on Tuesday, Feb 5 at 3 pm. We should go so that you can tell them that DOE MACSYMA has ALREADY done the same thing in NIL. Actually, it should be an opportunity to meet the guy from New Zealand who is talking, and see what they intend to do vis a vis DOE-MACSYMA. From: LPH 3-FEB-1985 14:30 To: GJC Subj: RE: talk at symbolics excellent. it's best to hit them over the head with something we did awhile back, especially at their own meeting. but we'll have to be polite about it. i've got [share*] (except some [sharem] stuff) in shape, with current documentation, and will now work on [ode] to get the demos in shape. then wrap up the newsletter. you can also wonder about a new bug: doing a writefile/closefile combo inside a file which is batched causes the c-lines' contents to not appear in the output; but a writefile/closefile done before/after batching a file does get the output correct, except for possibly line numbering. sigh. just can't win sometimes. From: JER 4-FEB-1985 09:22 To: ALL_USERS Subj: NIL Manuals Finally, we have 10 copies. Come and get one, if you need it. Janet From: MHD 4-FEB-1985 20:40 To: GJC Subj: RE: one week. ok. by the way, is there any reason why format in NIL can't take a string w/fill-pointer as documented in Lisp Machine 'ual? From: MHD 4-FEB-1985 20:44 To: GJC Subj: format and kermit-frame Last question re string as destination. Also, check out hand-down and hand-up commands at top-level in kermit. (60 line terminal!). -mark From: LPH 5-FEB-1985 09:56 To: GJC Subj: today i hear that RZ&JM's course is TTh 11-12:30 in 1-2xx (call mit info at 253-1000) I have a meeting at 11 myself. I'll be back here after i make the class, if possible, and will call you around 2:15 to see that you get away in time to go to the symb meeting at 3pm. I left a message for dennis stevens last night, and will be back here after the talk at symb to see if he returns the call. i'm also expecting that Grant will call. i also have to get the bibliography for RP, which will entail a little running around. From: DAWNA 5-FEB-1985 11:22 To: GJC Subj: Comm. Mort. Poodles, The guy is coming form Commonwealth Mortgage then and you know it takes a while. Do you really have to go? *** D From: DAWNA 5-FEB-1985 11:25 To: GJC Subj: Comm. Mort. He's coming at 2:00PM. I need for you to make out a check for the fee. Can you put some money in my account or start up the joint account soon, because it dosen't look like I can afford to pay the closing fee. I don't need to touch the money, I just think it would look good. ****Dawna From: LPH 5-FEB-1985 13:45 To: GJC Subj: course Sandy Hvatum called and left a message that the course was cancelled, and that he would try to get more info From: PJR 5-FEB-1985 15:38 To: GJC Subj: RE: vax usage fine..........what do you suggest??? From: MHD 6-FEB-1985 13:51 To: GJC Subj: 60 line terminal Any idea how to get a sixty line heath terminal from steve? -mark From: LPH 6-FEB-1985 14:36 To: GJC Subj: RE: RHB benchmark as soon as practicable. if you have a macsyma band on a machine, i can run the thing myself. i can come over at anytime. i ran into WGD in the mit library and he said that RNB is consulting for symb in fortran (i'm not sure if it is the macsyma group). i later ran into CWH in the hallway and he said that we should take displayedit, soup it up, use a mouse with it on a lispmachine, and advertise all of the enhancements! From: LPH 6-FEB-1985 16:25 To: GJC Subj: RE: RHB benchmark The file [lph]rhb.ben calls the interpreted, translated, or compiled versions of rhb.mc, .lsp, or .vas depending on what you request. A writefile must be wrapped around the whole thing in order to verift the answers and the times. When should I pop over? From: LPH 6-FEB-1985 19:57 To: GJC Subj: RE: macsyma and rhb Call me at 492-6079 when you want to try it out. I'm up mornings these days. I talked to Dennis Stevens, and he is working on the foreign distributors to see if they can get the machine, and i'll try to get the code. Spoke with Grant about Kevin's NAGLINK (he liked it a lot), about the Convex C-1 (it's CRAY/4 for .675 M$ with 128 Meg Ram and 4GB addressable virtual memory), and New Scratchpad (public domain, so algorithms for algebraic case of Risch might be there for the converting to MACSYMA). Spoke with JIM, and he wants me out at Los Alamos around March 12 for a week, so I'll go to LA on 2/25 or so, then maybe to Grant, to Las Vegas, and then to JIM before returning to Cambridge around March 18. Then to Austria for April 1-3 after a rest. Is it possible to get JIM a LAMBDA with DOE-MAX while I am in Los Alamos? That would give them an opportunity to judge the machine's performance compared to 3600, as well as give me a super-working macsyma to show off with to their theory groups - otherwise, I may get stuck with an old Vaxima that doesn't run the stuff I want to show (plotting, translation, SHARE stuff...) So any chance of a loaner at that site? From: MHD 6-FEB-1985 20:37 To: GJC Subj: RE: remote server mode is working now. Great. How do you make logical names? I want one for lam5. -mark From: LPH 7-FEB-1985 08:29 To: GJC Subj: JIM & LAMBDA Very good. As you can see, I'm up at a "normal" (ugh!) hour today, so I'll head into MIT sometime, but I would like to see the RHB benchmark in person if you do have a chance to run it. Thanks From: DP 7-FEB-1985 10:28 To: GJC Subj: chaos well put sent a 150847 char file completely intact. get retrieved the same file, adding a carrige return at the end. (I don't know which end is responsible for the addition.) please either describe fully (ior do yourself) the additions necessary to have it work with the rest of world. thing on the list: fix put server to do other than single char i/o add server for delete, and directory. possibly fix put and get user programs to do other than single char input, as the world is currently slower than lam3 at actual transfer. (if it runs at faster than 6k cps I will be suprised.. bandwith shuld be closer to 50-80k cps) most of that is spent doing :tyo's as the processor spends all it's time in run state jeff From: MRC 7-FEB-1985 15:15 To: GJC,MRC Subj: Streams (again) George, Can you tell me the filename and version number of the Streams example that will go out on the tape with release 2? This is so that I can put the matching example back in the document in hardcopy as you wanted. Thanks, Meryl From: LMC 7-FEB-1985 15:18 To: ALL_USERS Subj: wish-lispm mailing list on cap There is a new mailing list on Capricorn, which we can use as a wish-list for things we would like to see in the future on the Lisp Machine. Even if you think the wish is already known, or you think it will never happen anyway, send it to wish-lispm. Wishes sometimes even come true. Linda From: LMC 7-FEB-1985 15:18 To: ALL_USERS Subj: wish-lispm mailing list on cap There is a new mailing list on Capricorn, which we can use as a wish-list for things we would like to see in the future on the Lisp Machine. Even if you think the wish is already known, or you think it will never happen anyway, send it to wish-lispm. Wishes sometimes even come true. Linda From: PJR 7-FEB-1985 15:36 To: JEFF,GEORGE Subj: modem lines. Could someone explain the way the modem lines work. i need to dial into another vax off site.???????? thanks ....PAUL From: MHD 7-FEB-1985 21:55 To: GJC Subj: kermit From: PUTNAM 6-FEB-1985 21:51 To: MHD Subj: kermit More bothersome Kermit questions: In the login server REPL, how do you stop output from a runaway form (i.e., what's the equivalent of ^-ABORT)? Is there any way to send supdup I/O to the remotely logged in [sic]? How do you cancel the login server? Could you add a local echo option to the list of Network commands? robert From: WILDE 8-FEB-1985 08:31 To: GJC Subj: ARPA mail Found this staring at me from my bulletin board: mitccc!foo%bar.ARPA%mit-mc for sending ARPA mail from CAP to foo@bar. better late than never?? From: DP 9-FEB-1985 09:19 To: GJC Subj: chaos has been moved to the dba0:[chaos] directory. still to move are the c include files, and the c utilities (from lmi$disk) the ncp runs detached, and there is a command file to launch it woven into the system startup. put is still slow. the debug timers are still in place. everything has been re-compiled, and new ling files (based on an object library with all the chaos cruft in it, so the moby link statements are a thing of the past I left the ncp running jeff ps in the get user stuff, try get-file j. From: DP 9-FEB-1985 09:29 To: GJC Subj: ... Sanford Camera (mass ave arlington, just past fire station) did a winning job fixing my XA.. they were even cheap and fast... (2 days turnaround) putfil is faster (lispm end) but still slow (vax end) jeff From: LPH 9-FEB-1985 14:07 To: GJC Subj: RE: bench marks only about 10% faster in the compiled code. not even as good as interpreted code on the 3600... also, i was wring about multics time for 1000!$ - it is 1.35 secs vs 1.42 on LAMBDA, after running a few times. I have to be back by 5pm to go to dinner, but i could come over and hack if you like. From: LPH 9-FEB-1985 16:35 To: GJC Subj: RE: bench marks since even the compiled timings are worse than the 3600 interpreted ones, that doesn't seem to good... From: LPH 9-FEB-1985 16:54 To: GJC Subj: RE: how worse The timings are on the sheet of paper stuck in the muc3 proc. and is right next to where i was doing the work. the file lm:tmp;rhb.mwXXX is probably where i left them on-line. probably some 10-30% worse, on the whole From: LPH 9-FEB-1985 17:12 To: GJC Subj: timings i'll be around on sunday all day, so we can discuss the details at your convenience. From: LPH 11-FEB-1985 09:48 To: GJC Subj: 9083 could the modem be returned on that line so i can kermit? thanks. i'll stop by today sometime if i can't use kermit by the late afternoon From: JER 11-FEB-1985 17:00 To: GJC Subj: your kermit doc I had a bunch of stuff on my mind when you mentioned your kermit doc to me earlier today. So, what is your document? Is it something we should turn into a regular customer document? We don't actually sell Kermit to customers, do we? What's the LMI policy on supporting Kermit? Or who else should I ask these questions? Thanks, Janet From: DP 13-FEB-1985 06:04 To: GJC,MWT,PJR Subj: tcp and vax config Vms tcp is running using an interlan board (the spare that keiu had) the 3 com board has been removed. vms tcp was able to reach the unix vax. vms chaos is using the original interlan board. all the code for using the 3com board has been diked out. It was not easily possible to get tcp to use the 3com, so I punted rather than fix it. vms chaos definitely comes up when the system is booted... (there was a slight problem with a debugger linked version in the run directory) the second interlan board is at csr %o765000 vec %o540 From: LPH 14-FEB-1985 22:32 To: GJC Subj: things I talked to Jeff Blacksin at DEC - he wanted DOE-MACSYMA info so that he could do some work using it. He said that MicroVaxII will not be available for a year, even to his group...but he would pass me on to marketing, which may shake loose some time on a I instead (if they don't insist we just buy it.) Dennis Stevens will get me in touch with Grapheal Friday at 1pm our time and he thinks that a 2x2 will be available for Linz. Please call me at 492-6079 about noon, or 253-2539 in the early am, to see what is up on this point. How did meeting with JIM go? From: LPH 16-FEB-1985 13:15 To: GJC Subj: hi how'd it go? From: DAWNA 16-FEB-1985 13:16 To: GJC Subj: terminal Brigette Carter called from Radio Shack about the terminal 354-7694. You poodle! From: LPH 16-FEB-1985 13:17 To: GJC Subj: sounds ok looks like Bender & Guralnik's type of work. shall i come over now? From: DAWNA 16-FEB-1985 13:17 To: GJC Subj: Wooff How do you send messages like that poodle? From: DAWNA 16-FEB-1985 13:19 To: GJC Subj: Drama shop Did you call? From: DAWNA 16-FEB-1985 13:37 To: GJC Subj: botex Anytime I try to do anything in Botex I get the message, Previous logical name assignment replaced, how come poodle? From: LPH 17-FEB-1985 13:35 To: GJC Subj: amort The problem with MSIZE stuff is not a bug, but just the result of incorrect programming methodology. The generation of the proper environment in which to avoid meta-programs looks like something worth developing as a subject for the Newsletter - it will surely be one of the most sophisticated uses of the programming language. From: MRC 18-FEB-1985 13:59 To: GJC,MRC Subj: final version of online example for streams release 2. I need to know the name and version number so that I can include a hardcopy. I'm finishing this, this week and all I need from you is a filename. -----Meryl From: JER 19-FEB-1985 09:37 To: GJC Subj: RE: tiger and kermit documents Thanks for the info. Yes, George Colpitts and Bob Putnam of CS will review the manuals and make sure everything works. So, just let me know when you make additions. I'll make sure they get in the manual. Boy, you sure looked nice yesterday! From: MRC 19-FEB-1985 10:03 To: GJC Subj: RE: final version of online example Thank you. mrc From: WILDE 19-FEB-1985 15:49 To: GJC Subj: same time, same place encore performance culture lecture scheduled for next thursday, feb. 28 please confirm and let me know if you need anything. thanks again. From: LPH 19-FEB-1985 21:33 To: GJC Subj: [lph]rjf.txt is my proposed draft of the SIGSAM info on DOE-MAX. please give me your comments asap. thanks. From: LPH 21-FEB-1985 20:02 To: GJC Subj: RE: incredible winnage Great! Then we can run SLATEC right away...So Grant will love it. Just talked with JIM. He is trying to get me 25k (includes travel) for 60 days work (some here, some in LANL, at 300/day or so), so that sounds great to me. How about the RJF.TXT note for SIGSAM? On another topic, JIM is concerned about the appearance of a LAMBDA being strictly coincident with my appearance there. Can the LAMBDA be left for a few weeks to enable them to check it out fully? Or can the DOE-MAX version run on the TI EXPLORER they have? Or can I just get our SHARE sources on a tape to dump onto their 3600 MACSYMA's? We need to talk to JIM about this asap to clear it up... From: DAWNA 22-FEB-1985 10:52 To: GJC Subj: il There is supposedly a graphics program call il at MIT. Is there any way we can get it over here? ***Dawna From: LPH 22-FEB-1985 13:32 To: GJC Subj: RE: lambda JIM's problem is that it looks like the LAMBDA and I come as a unit, which is advertising for LMI rather than meeting the needs of LANL. So it is a matter of appearances for propriety's and fairness' sake that the LAMBDA not be tied directly to me. I'll take out "lexically scoped" and add in a sentence on being able to call other compiled subroutines. From: DAWNA 22-FEB-1985 15:16 To: GJC Subj: il's author RPK told me that he thinks the author is Duarte Morais or something like that at LCSI. ***D From: LPH 22-FEB-1985 20:14 To: GJC Subj: RE: lambda If the LAMBDA is not available during my stay, then perhaps we can try loading the tape onto their EXPLORER or their 3600's? I would like to be able to show off some of my work on SHARE, including the compiled things... As a front-end to super-computers, the LAMBDA has it over the other LISPM's, getting a machine there is imperative, but the time schedule need not fit my stay there... From: LPH 23-FEB-1985 15:25 To: GJC Subj: RE: lambda Good. Then a tape should be prepared that I can use. Have you made your travel plans yet? We can probably get a good fare for your trip this weekend, but then it will go $100's higher during the week. Grant was not in the past week, but is expected Monday. I also need some travel advance money and so do you. Will you be in for awhile? From: LPH 24-FEB-1985 02:49 To: GJC Subj: [share] i just cleaned out the .txt, .trl, and .dmo files that were not needed and the remaining ones should hang around still. now there are only 7,163 blocks in [share*]... can ttyintfun from [jpg]suprv be #+NIL'ed ? From: LPH 24-FEB-1985 03:34 To: GJC Subj: remaining bugs no %% in control-A break matrix([a,b],[c,d])^^-1,detout,ratmx:true; errors out in rational functions package (it always did, not a new bug based on sat nite's work) editor does not yank last c-line, and myvc1$ myvc2$ gives a useless error when the editor is used. array lossage prevents ndiffq from working. batching a file which starts its own write/closefile causes the clines to not appear in the output file. [demo]runge.dem is the specific file for the array lossage above, and [share]array shows rearray loss. that's about it...nearly everything else is satisfactory for the moment. From: LPH 24-FEB-1985 04:43 To: GJC Subj: RJF liked the submission for SIGSAM, and asked about the following mis-feature: integrate('diff(y,x),x) is not y, but the noun form of the integral. this might be real trivial to fix in SININT, say, by checking the main operator of the integrand to see if it is EQ to '(DERIVATIVE SIMP) and if so, reduce the order of the derivative by 1 or return the first arg of the der if order is one...i'll look into this sunday afternoon From: LPH 24-FEB-1985 13:20 To: GJC Subj: RE: integrate('diff(y,x),x) I just fixed this bug. it was trivial in sum-of-intsp, and is patched in 8p23. so is the ilt([laplace(f(x),x,s)=...,laplace(g(x),x,s)=...,...],s,x) which occurs in the [demo]laplac.dem From: LPH 24-FEB-1985 13:23 To: GJC Subj: RE: ttyintfun Ok. this is for use with such as romberg, quanc8, diffeq, etc that say where they are at some point in time when you give the interrupt. since you have to type a control-c anyway on the vax, typing A for a break, checking what you want, and then exit; to restart the computation is only a minor loss compared to real bugs... From: LPH 24-FEB-1985 13:29 To: GJC Subj: rearray is actually only partly broken. you can make a 1x10 array become a 2x5 array, but not a 10 array, i.e., you can't change the number of dimensions. and fillarray seems to put 0's where 0.0's would be appropriate in some cases. these things need careful checking out... From: LPH 24-FEB-1985 14:02 To: GJC Subj: laplace I figured that I could as easily make both ILT and LAPLACE map over their first args if they are lists, so that is now in the source. I won't re-compile the patch until I look over a few other things... From: LPH 24-FEB-1985 14:04 To: GJC Subj: operators The idea of OPERATORS in laplace, diff, integrate, etc. is a good one. Doohovsky (sp?) did some work in that area awhile back. Perhaps JPG has achieved some of this functionality in his work at GTE. From: LPH 24-FEB-1985 16:53 To: GJC Subj: taylor(sum...) is not reliable. [lph]sumtay.mw is such a case From: LPH 25-FEB-1985 01:38 To: GJC,LPH Subj: taylor(sum(n*(n-1)*a[n]*x^n,n,0,inf),x,0,2)-->0+... This bug is due to [rat]hayat function TSPRSUM executing the (return ans) clause in the last if clause at the bottom of the function, even when it should sometimes go to the psplus clause instead. this is rather a pain... and may be the root of a lot of Taylor's other problems. From: WILDE 25-FEB-1985 08:44 To: GJC Subj: ed-hacks would appreciate your checking out the function of the local net for the ed machines. While it works OK on Release 1.2, am getting erros such as "lmi-godot is not a known host" on 102.84 (while in si:parse-host). Also needs to have the package ed-hack created from the debugger -- changed the mode line, but am probably missing something obvious. The original code you wrote is in "lam3:lmc;ed-hack.lisp"). Please keep me posted. Thanks. From: MWT 25-FEB-1985 14:11 To: ALL_USERS Subj: tcp status Tcp is up and running again, and it is much more reliable than before. However, it still flakes out occasionally, or goes into 'molasses mode'. Feel free to reboot the bridge as often as is necessary -- the machine is in my office, and the system console is on the auxil port of the Z-29 in Kieu's office. Mark From: MWT 25-FEB-1985 14:11 To: ALL_USERS Subj: tcp status Tcp is up and running again, and it is much more reliable than before. However, it still flakes out occasionally, or goes into 'molasses mode'. Feel free to reboot the bridge as often as is necessary -- the machine is in my office, and the system console is on the auxil port of the Z-29 in Kieu's office. Mark From: LPH 26-FEB-1985 03:12 To: GJC Subj: dessert My apartment mates made a chocolate fondant, an elegant dessert, for my birthday (wednesday). Why don't you and Dawna stop by after lunch for a slice. From: GARY 26-FEB-1985 12:16 To: GJC Subj: LLNL Trip So glad to hear that you will be coming to the West Coast next week. All the arrangements have bee made to have the Lambda at Livermore by Monday morning. If you arrive over the weekend does it make any sense to do some work on the Lambda here in the office? Let me know and I will make plans to let you in. I really haven't heard anything more about Jim Odell getting a machine. I will have to call him myself instead of letting Gruber do anything about it. It seems like I've got a winning way to dial up a Lambda remotely. The only problem is what to do when thrown into the error-handler. I tried pre-parsing but that it only catching some errors. Alex claims that you fixed the problem in that hack I gave him but he may have been wrong. It sure would be nice to figure out how to return a message rather than entering the debugger. Hope all is going well for you. Let me know if want a ride from the airport or something ----Gary K. From: LPH 26-FEB-1985 12:22 To: GJC Subj: [aareadme]8p23.txt I will be showing up at LLNL myself on Wednesday next week while you are there, and will work on the file when i get there. I have postponed my trip to see JIM until mid-April to take some pressure off him. I'l be back here in two weeks from tomorrow. I will go to Las Vegas the end of next week, and go to see Grant again Monday the 11th, before returning to Boston via LA. I will hit San Diego people at GA and Jaycor (they get the Newsletter) on Monday next week as well, so quite a romp. Stop by for chocolate fondant - give me a call and come over. From: LPH 26-FEB-1985 13:47 To: GJC Subj: PicTel They want you to call them (1-535-7700) and are interested in your coming out as soon as reasonably possible to help them in making a phone-back modem (prevent unauthorized use). Talk to Brian Hinman or Tom Spalding. From: LPH 26-FEB-1985 15:15 To: GJC,DAWNA Subj: chocolate fondant You both seem to be out at the moment (or at least not logged in) so I am going out for awhile and won't be available until this evening if you want a bite of this treat. Call me after 6 or so... From: LPH 26-FEB-1985 16:13 To: GJC Subj: tonight saw Dawna in Harvard Sq. She says that after 9:30 is ok with her, so just call me when you want to come over. From: MWT 27-FEB-1985 04:04 To: ALL_USERS Subj: Moving files between a Lambda and the VAX Can now be done directly from the Lambda, without having to telnet to the bridge, run FTP, etcetera. Just load the files "lam3:tcp-ip;bridge-2.qfasl" and "lam3:tcp-ip;tcp-bridge.hosts". (If the Lambda's running Release 1, load bridge-1.qfasl instead of bridge-2.qfasl). This will define two new hosts called "VMS" and "ANGEL", and you can read/write/edit files on those hosts just like any other machine (more or less). for example: (fs:copy-file "vms:user$disk:[smith]stuff.lsp" "lam3:smith;stuff.lisp") does the right thing... Dired doesn't work though, and there's a couple other restrictions -- see lam3:tcp-ip;tcp-bridge.doc /Mark From: MWT 27-FEB-1985 04:04 To: ALL_USERS Subj: Moving files between a Lambda and the VAX Can now be done directly from the Lambda, without having to telnet to the bridge, run FTP, etcetera. Just load the files "lam3:tcp-ip;bridge-2.qfasl" and "lam3:tcp-ip;tcp-bridge.hosts". (If the Lambda's running Release 1, load bridge-1.qfasl instead of bridge-2.qfasl). This will define two new hosts called "VMS" and "ANGEL", and you can read/write/edit files on those hosts just like any other machine (more or less). for example: (fs:copy-file "vms:user$disk:[smith]stuff.lsp" "lam3:smith;stuff.lisp") does the right thing... Dired doesn't work though, and there's a couple other restrictions -- see lam3:tcp-ip;tcp-bridge.doc /Mark From: LPH 27-FEB-1985 07:10 To: GJC Subj: max$disk:[lph] I put the promised programs (MUC3 promise) there, so be sure that they get on the tape that Grant will release. CUnextWednesday! From: MHD 27-FEB-1985 22:17 To: GJC Subj: get a file I need to get mcdoc from lam3? Is there anyway now? -mark From: MHD 27-FEB-1985 23:10 To: GJC Subj: phone me phone me. -mhd From: MHD 27-FEB-1985 23:15 To: GJC Subj: RE: up yours Thanks, and shove it. From: SLOAT 28-FEB-1985 00:13 To: GJC Subj: Microcode Compatibility George, Have you any speculations as to whether a System 98 Band would run with a system 99 Microcode? We currently run with Mic 536. The increased virtual mem would be a real winner. Our internally distributed bands are so full of "TOOLKITS" that the distributed bands are already past the GCable point. How do you like your explorer?? Regards, Jay Sloat From: DP 28-FEB-1985 07:14 To: GJC,RPK Subj: chaos stuff the put code on the lispm always adds an extra (higher pkt numbered) EOF packet when sending. it then waits for this pkt (despite getting a CLS pkt back from the server) this is the long net finish (it won't ever get ack'd, as the server process had quit and wasn't going to read it... (and until that pkt is read , by the user's proces, it cant go away) ca the getfile server now sends :length as part of the atribue/truename packet. next - version numbers.... I re-compiled the do-getfil +assorted other things, they don't work anymore (no changes, just tried another compile, got lots of errors, etc.. the vax getfil server is still running, just not rpk's lispm code. a minus -1 packet number (which you aoocaionaly see on rfc's and opn packets from the lambda) caused the vms stuff to not fill in a few fileds. this may be why microcode stuff doesn't work...(tho there wasn't a problem with the hostat code...) From: DP 28-FEB-1985 07:16 To: RPK,GJC Subj: packet space dump there is a dump of the headers (and a key page to go with it ) on my desk, if you are interested in seeing at the bit level jrff From: RPK 28-FEB-1985 10:47 To: DP,GJC Subj: DO-GETFIL Yeah, I was in the middle of changing the error signalling stuff. I will fix it before you get in; if the NCP is running, I'll test it. From: SWRS 28-FEB-1985 12:56 To: GJC Subj: RE: printer documents Thanks! From: DP 1-MAR-1985 07:45 To: GJC Subj: RE: ncp yes, but Dan Thomas told me not to run it during the day if there was a chance it would crash the machine... (he want's the system available for the documentation people....) so until interlan comes through.. (harvey says they now have a 2-3 week backorder situation on the driver kits... waiting for tu58's...) I can't trust it while the rest of the world it hacking... jeff From: DP 1-MAR-1985 07:51 To: RPK,GJC Subj: state of chaos code we have version numbers now... the putfil bug that adds 2k of zeros to the output file when it hits the end of a packet has re-appeared (actualy not a copy problem as much as it is a simple miscalculation of length length of destination string by chaos_in_string) jeff From: RPK 4-MAR-1985 15:39 To: GJC, DP Subj: Testing of VMSFIL access It tried it out a little this afternoon from LAM15B, which is a 2x2 running Release 2 (it's a Process Control machine). Anyway, it seems that even a window size of 6 can screw things up. I could read in my LOGIN.COM, but things would hang at 82% of [RPK.DOC]MC.TEX. A window size of five, however, always seems to work, but slows the transfer down notably. The NCP's flow control code might be at fault here... I also tried writing a file -- which seemed to work from the Lambda side OK. However, when I tried TYPEing the file (USER$DISK:[RPK]VMSFIL.LSP;6) from the VAX, I got a RMS-W-RTB error; a 2048 byte record was too big for the user's buffer. I didn't try reading the read file from the Lambda -- I'll probably give that a shot when the machine frees up again. Todos for the file servers: Login, so that people follow VMS access controls, and leave their UID's on the files they write. Options for opening a file -- the only immediately useful one I can think of is binary. Other things include appending and a preallocation size argument. Error codes for file errors. The QFILE protocol defines a whole bunch of three-character error codes that this protocol should use as well. (This save a lot of Lisp programming...) Another server that would be nice to implement is a MAIL server.... From: DP 6-MAR-1985 04:50 To: GJC Subj: chaos progress From: DP 6-MAR-1985 04:48 To: DP Subj: chaos progress From: DP 6-MAR-1985 04:47 To: RPK Subj: chaos current state.... putfil seems to work. (no more 2k byte records) getfil ^x^f from emacs successfuly got a 47kbyte file. I was not able to write it back as I kept getting errors (unkown operation :open) when I typed ^x^w or ^x^s will change getfil to return fnf (which case would you like... seems like the document has things in upper case. started looking at access validation stuff.. ) packet text ("filename.ext" :atrib :atrib :atrib) "user" "password" or maybe ("filename.exe" :user "username" :password "pass" :atrib ...) (the second would be more short term work to parse, but a parser for that form is needed eventually) the putfil problem was my forgetting to to return a status code on success (I did return one If I failed) in pkt_from_read (called by chaos_in_string) I also added the suspenders that it always calculates a string length (prev it left the length field of the descriptor alone if it errored... it now forces a zero read length, as nothing can be counted on...) jeff From: DP 6-MAR-1985 05:01 To: GJC,DP,RPK Subj: top of fixit list - chaos slow net finish - putfil window size sensitivity - getfil access control - file ownership problems first then passwords. qfasl/binary mode - put jeff From: DP 6-MAR-1985 05:14 To: GJC Subj: chaos size the chaos ncp currently takes 6+ minuites of cpu time to compile and link. jeff From: RPK 7-MAR-1985 14:09 To: GJC,DG,KHS,@[SWRS]DOCGROUP Subj: BoTeX 0.16 BoTeX version 0.16 will be installed at 2:30 today (7 March). There are no compatible changes. You can read about the new features in [RPK.DOC.TEX]BOTEX016.TXT. Bugs to me. From: RPK 7-MAR-1985 14:27 To: GJC,KHS,DG,@[SWRS]DOCGROUP Subj: Ooops I meant there were no @i(incompatible) changes in version 0.16. Sorry to cause a panic... From: DP 8-MAR-1985 01:15 To: RPK,GJC,DP Subj: chaos crashes problem with ncp just going away at random (takeing vms with it when not debug version) was due to the io check stuff considering the transmit with retry (collision) as a fatal error.... fixed now. I was able to get ~500 kb of data without a single error jeff From: LMC 8-MAR-1985 14:48 To: ALL_USERS Subj: Using TCP/IP from the Lisp Machine Please try using TCP/IP from the Lisp Machine via a UNIX bridge. We'd like to get as much use as possible of this software under the Beta Release 2.0 software (System version 102.92, microcode 753). Instructions on use can be found in lam3:tcp-ip;tcp-bridge.doc. I printed that file, made several copies, and put them on the counter next to the mail boxes in the common area. I also have some extra copies in my office. Comments, suggestions and bug reports to Mark Terpin. Linda From: LMC 8-MAR-1985 14:48 To: ALL_USERS Subj: Using TCP/IP from the Lisp Machine Please try using TCP/IP from the Lisp Machine via a UNIX bridge. We'd like to get as much use as possible of this software under the Beta Release 2.0 software (System version 102.92, microcode 753). Instructions on use can be found in lam3:tcp-ip;tcp-bridge.doc. I printed that file, made several copies, and put them on the counter next to the mail boxes in the common area. I also have some extra copies in my office. Comments, suggestions and bug reports to Mark Terpin. Linda From: DP 11-MAR-1985 10:57 To: GJC Subj: sizes brand j macro 5795 lines 162176 chars c 3315 lines 96619 chars total 9110 lines 258795 chars brand x source 26390 lines 829125 chars com 2194 lines 72769 chars total 28548 lines 331319 chars jeff From: JMTURN 11-MAR-1985 16:14 To: ALL_USERS Subj: CAP The good old CAP command is back, which gives you a connection to Capricorn. There's only 1 serial line, so it may come back with "DEVICE ALL READY IN USE", or some other friendly message. Type Control-Shift-6 followed by Q to exit. Sorry it's only 1200 baud, but the Vax gets unhappy if you try to go any faster. James From: JMTURN 11-MAR-1985 16:14 To: ALL_USERS Subj: CAP The good old CAP command is back, which gives you a connection to Capricorn. There's only 1 serial line, so it may come back with "DEVICE ALL READY IN USE", or some other friendly message. Type Control-Shift-6 followed by Q to exit. Sorry it's only 1200 baud, but the Vax gets unhappy if you try to go any faster. James From: SLOAT 13-MAR-1985 19:24 To: GJC Subj: AP for Lambdas?? George, While mulling over the Release 1.2 Sources I saw a few references to a Multibus Array Processor. Is this working and is it available as a product yet?? Several TI'rs were interested in the Particulars (ie. Cost,Performance,memory requirements, ect.). Regards, Jay Sloat From: LPH 13-MAR-1985 21:21 To: GJC Subj: i'm back in town now. when is good to hack lambdas? From: LPH 13-MAR-1985 21:23 To: GJC Subj: RE: time I'll be over in ten mins... From: LPH 14-MAR-1985 01:20 To: GJC Subj: through the M's, the errors are not too many...the file lam3:lph;bugs.lam and the sheet have the new specifics... i'll be by tomorrow when i am more alert, but call me in the morning if the machine is freely available and i'll get back to work on it. From: MCC 14-MAR-1985 17:19 To: GJC Subj: RE: arpanet address I gather that your message was for Katie. You should be able to mail to her at. rotzell@mcc.arpa - Rob From: JMTURN 14-MAR-1985 18:19 To: ALL_USERS Subj: Capricorn line Due to some techincal problems with the line, as well as a generic problem with the security of the dialups (which has to be solved before any Capricorn interface can return), the Vax->Capricorn line has been removed. From: LPH 14-MAR-1985 20:38 To: GJC Subj: when is a good time to stop by again From: LPH 14-MAR-1985 20:39 To: GJC Subj: when is a good time to stop by again From: LPH 14-MAR-1985 21:41 To: GJC Subj: hi anything new today? i have heard from PicTel that 50 dollars per hour rather than 75 should be used. is that what you intended? From: LPH 14-MAR-1985 22:02 To: GJC,LPH Subj: taylor bug the commented out code in tsprsum, when put back in place, fixes the taylor(sum(n*(n-1)*a[n]*x^n,n,0,inf),x,0,3) bug, so i am adding it to 8.23 and adding the atom check in my last fix for integrate in sum-of-intsp, and then i will re-compile 8p23.lsp From: LPH 15-MAR-1985 11:28 To: GJC Subj: finish testing when is a good time to get the rest of the items tested out? i'm available at all times, basically. From: LPH 15-MAR-1985 19:44 To: GJC Subj: RE: do not recompile 8.23 - use 8.24 sorry, too late. should be recoverable by a diff on 8p23.lsp versions to make into 8.24 From: LPH 15-MAR-1985 22:46 To: GJC Subj: lispm doe-max has its problems, but it seems to work on the .mc files in share* stuff that I tried. %th is now fixed, it seems. I want to kill Grant's re-definitions of collapse (in optimize), but can't seem to manage. call me saturday afternoon sometime and we can straighten that out so my demos run ok. then we'll test out some of the newsletter items. From: MHD 17-MAR-1985 01:54 To: GJC Subj: RE: kermit Looking at the file I happened to have on [mhd.kermit2]proto.lsp I can see that there's a do missing from that form. I don't think there's an implicit progn with as in loop. There should be a do right after the as line. This should not be getting past our compiler by the way, I don't think. -mark From: LPH 17-MAR-1985 02:37 To: GJC Subj: int.dem on lambda I know how to make it better now, so be sure to get me in tomorrow afternoon to work it over. We then need a tape for NESC as well. From: LPH 17-MAR-1985 22:35 To: GJC Subj: release notes take a look at user$disk:[lph]lambda.rel before you leave, and make changes as appropriate while you are creating the changes tape to take with you. I'll fed-ex the tape in the afternoon. After I let Grant know the status. Have a good trip, and keep me posted. From: LPH 18-MAR-1985 23:33 To: GJC Subj: gif He logs into anl-mcs, a unix vax, and copies off files that are for vaxima, generally, along with some things like zrpoly (from imsl) and gjc;cracko (for adk from '80). he then ftp's these all to xx using myk.gst as login; and deletes the files on mc. From: LPH 19-MAR-1985 16:41 To: GJC Subj: tex ran fine, after I got [jmturn]tex.com copied over [it is the only one that I had access to, since you had your protection set for your use when you made my copy]. the tape went out monday afternoon. Grant says that he has received confirmation that I can take the tape for VAX VMS to Linz. From: DP 19-MAR-1985 23:42 To: GJC Subj: stdio jmturns stdio was missing the atoi (asci to integer) routine. I copied the code from K+R into the chaos private copy of stdio. jeff From: DP 20-MAR-1985 07:58 To: RPK,GJC Subj: chaos the gerfile server now takes an atribute list for the filename packet. the things that can be specified can be found in the file chaos$src:chaclib.c getfile now sends the creation,revision,etc.. dates back in it's truename packet. you can send username and password to getfile. it still ignores them, but it wont be upset to find them there. putfil didn't get changed. tonight I will do so, and star it toward recognising usernames. syntax: ("device:[direct]filename.ext" :username "foo" :password "bar") abreviations are not recognized. case does not matter. it ignores spaces, tabs, newlines. all programs (so far) accept filenames, and return truenames. all dates are in vms standard format. I expect to parse vms format and universal time on input. new personal cpu time record: 1:18:30.63 jeff From: DP 21-MAR-1985 08:34 To: RPK,GJC Subj: chaos state putfile now takes plists. there is a new property available :blocksize this is used with binary mode transfer to set the fixed record size. (I suppose it will also be sueful when we do magtapes) I am currently fixing putfile to cope with binary files. I looked at the username and validataion stuff. gross. ugly. requires locking pages and shuffling things around at ipl$_synch. performing unnatural acts with the jib, etc. on the sigtape, there was this program called become, that did all that. unfortunately I don't have the source... (I don't know if it came with it and I deleted to save space or what.... I don't have the original tape... I have asked of info-vax if anyone has such a thing. failing that I will type from the fiche (loginout). this is an obvious place of a special jeff From: DP 21-MAR-1985 10:16 To: RPK,GJC Subj: chaos - putfile putfile now correctly processes the :submit :print and :delete options jeff From: DG 21-MAR-1985 11:38 To: GJC Subj: RE: make-distribution-tape Sorry... I thought everyone knew that the version on LAM3 was the one to use... the version I've been hacking is the one on LAMH... but I'll move it to a secret place and bash it there... Sorry to cause lossage... there's just alot more I want to do with it and there have been so many distractions. I'll check the version on LAM3 and make sure it works, and back that version up to dg.backup; on LAMH too. I'll put the development version somewhere else. Go ahead and extend/fix it as much as you wish... i'll mark the current versions (DONT-REAP or something) and compare/merge them into the new version. -dave From: COLPITTS 21-MAR-1985 16:09 To: GJC,MHD,PUTNAM Subj: kermit server on unix is a kermit server on unix possible From: LPH 21-MAR-1985 23:12 To: GJC Subj: nesc got our tape. they want a listing of the files and a confirmation of the number of files to ensure that they have 1348 files as a complete set. How did it go in washington? I'm very anxious to hear. I have a draft of the ISA paper in user$disk:[lph]isapap.txt. It needs a lot of work in the section on NIL linked to other codes, since I have 0 experience with this. There is a lot of formatting that I have to do on the final version, so please look at this right away, and call me as soon as possible to go over it. I have to have it done sometime friday afternoon or evening in order to paste it up and mail it out by express mail on saturday so that it can arrive on Monday (the deadline after I called and asked.) From: WILDE 22-MAR-1985 10:32 To: GJC Subj: benched Welcome back. Have a question for you re: benchmarks. Please come find me when you get a chance. /lorin From: LPH 23-MAR-1985 00:25 To: GJC Subj: isa paper is done and ready for paste-up. Dawna said I could use her drafting table to get it done right. Sounds good to me. How did the tape making go? Give me a call in the AM so that I can get all of this done. I will start working on the Linz paper, but need more input from you on the experience of doing the porting, and other status indicators. From: LPH 23-MAR-1985 11:31 To: GJC Subj: i'm up now. will call you around 12:45 or so. cusoon From: LPH 24-MAR-1985 02:59 To: GJC Subj: linz.txt is my start on the LINZ paper. needs a lot of work, since many of my comments are purely speculation on my part, and are intended to serve as an outline for you to correct the statements and add additional info. please call me in the morning or noon-time, as this is very important to get through. thanks. From: LPH 24-MAR-1985 20:05 To: GJC Subj: sandy hvatum at draper says that we might be able to use his VAX for me to practice doe-max build in exchange for my showing him the differences with their REX version. he will check on it tomorrow and let me know. From: LPH 25-MAR-1985 01:29 To: GJC Subj: linz.txt Please look it over again. I have a hardcopy output on multics already, and can pick it up in the morning. Now to work on the transparencies... From: LPH 25-MAR-1985 03:43 To: GJC Subj: transparencies are started in eurocal.txt. i will have a hardcopy off multics shortly. we then have to go over it and make final revisions for my 15 min talk. also need graphics output from LAMBDA in order to show it working, and also a DISPLAYEDIT session would be good. what about FORT? or just regular macsyma demo-type stuff? From: ALEX 25-MAR-1985 09:42 To: GJC Subj: RE: hi. thanx..... How do I get it started???? From: LPH 25-MAR-1985 12:29 To: GJC Subj: shoosmith i gave him giansiracusa's phone number From: LPH 26-MAR-1985 01:17 To: GJC Subj: vax11/785 Great news!!! DOE-MAX was within 10% of REX for standard kernel functions. Because I compile TRIGSIMP in [share]trgsmp, a problem ran 3 times faster in DOE-MAX than the LISP-save file REX allowed. I will thus be able to show that the VAX11/750 in Linz will be as fast as the 3600 on this problem, despite the intrinsic machine speed differential. That will bring home to people some real interesting facts. Get me up early so that we can square away all of the details that need to be finalized. Joe Oliveira will probably get me to the airport by 5 if we leave by 4, so the more time i have the better... From: LPH 26-MAR-1985 02:35 To: GJC Subj: vax tape needs [share]grob.* and [demo]great.dem from lambda tape, and user$disk:[lph]rhb.mc can be added to max$disk:[demo]rhbtime.dem or somesuch. i also need to correct the instructions, and get LMI's telex number. a screen dump of a plot, the display editor, and my fortran optimization demo [let me work on this one] should complement my presentation. Also the timing info for the RHB demo on the lambda, in lam3: or somesuch. i am getting the edited paper ready; any comments on the transparency presentation (which i am keeping short in order to be able to answer the inevitable questions). From: JB 27-MAR-1985 07:57 To: ALL_USERS Subj: tape information Just a note to let people know that if you initialize a tape, it will destroy all previous information on that tape. This is to clear up any confusion pertaining to the tape documentation and to prevent anyone from unknowingly destroying their tape information. jane b From: MHD 3-APR-1985 20:47 To: GJC Subj: nil (arglist 'apropos) --> nil (apropos "string") --> nil meta-. on string-search, pretty-print, grindef, and pp --> not found Great system. -mhd From: DP 4-APR-1985 03:03 To: GJC Subj: various you will be getting a call from walter richtman (paladian software) please say nice things about me... chaos login works with the exception of setting uic... (it sets your uic to [0,363]) I will look at it later... I typed in (sans comments as I was hurrying) the password encrypt code. I will proof it tommorow when I haven't been stareing at it for hours. the set host code deassigns all your process local logical names... (which doesn't make much difference as you have so few in the context of a subproces to a detached one...) but it gets the default device and directory correct now. (since I do the deassign from exec mode, even the "permanent" logical names get flushed....) jeff From: DP 4-APR-1985 19:04 To: GJC Subj: authorization... uic now works... chmkrnl does a pushal of the arglist, so set_uic was setting the uic to the address of the uic part of the uaf record... jeff From: LPH 5-APR-1985 05:11 To: GJC Subj: your mail on mc must have amused the recipients... From: DP 6-APR-1985 02:46 To: GJC Subj: exelan I loaded the driver, and it found the device. the driver is NOT loaded at boot time. to load, sys$manager:exload. there is a host table, and a com file to start their NCP (netload.com) in [sysmgr.exos]. I did not start their ncp... jeff From: LPH 6-APR-1985 10:03 To: GJC,LPH Subj: interpolate a MACSYMA-level definition is now in [share1]intpol.mc and the .usg file is updated.