Jump Over Left Menu
Quarterly Progress Report 1 - 31 March 1975
F R A Hopgood and C J Pavelin
6 April 1974
The FR80 Project Branch was augmented this quarter by the return of Wade Shaw on a three-year research appointment and the arrival of Bob Witty. There was a further increase in numbers at the beginning of April when Miss Madhu Kashap, in her first year at BruneI University, came as a sandwich student for six months.
There have been no changes in the staffing of the Communications and Systems Branch. The current Group structure is:
- Communications and Systems Branch
P E Bryant
- D A Hopkins (Secretary)
- C J Pavelin
- G W Robinson
- S R Perkins
- R J Waters
- D A Duce
- A Idris-Bay (Student)
- M D Fowler
- J D Thewlis
- D C Toll
- P E Bryant
- FR80 Project Branch
- F R A Hopgood
- J B Chamberlain - Secretary
- R E Thomas
- M F Chiu
- P A Dewar
- A H Francis
- J R Gallop
- M B Kashap
- J W E Lewis (Â½)
- J M Rushby
- W D Shaw
- R W Witty
- P M Nelson
- R Brandwood
- D V Ralphs
- A W Burraston
- F R A Hopgood
The FR80 operators, G P Jones, D J Daniel, B J Jeeves and B J Lennon were seconded to the Group at the beginning of April.
2. PROJECT CONTROL (PEB)
All projects, ranging from a few days to many months in length, have now been catalogued. This includes projects not started and some which may not be implemented. Time scales are also included and other useful information. This should help with forward planning and gives a good idea of the activity of the group. The responsibilities for items of Software and 1906A system directories have been listed. The work is not yet complete but should be useful for assessing progress and allocating resources.
Mark 8 of GEORGE was adopted after some difficulties. The load on the system has been extremely heavy and a new stricter budgeting/scheduling system will be introduced shortly. The quarter has seen general use of NUTS as more facilities are added.
There have been no other major new projects started. There is a suggestion that ACL should continue a project started at Birmingham to provide a FORTRAN post mortem dump facility. This is being investigated in conjunction with UIJPC. No documents are yet to hand. There is also the possibility of initiating a project for documentation production similar to IBM's ATS system. Discussion and appraisal of requirements with Resource Management are in hand. The possibility of providing a new large scale disc sorting utility may also be considered.
All software changes made by the group are now listed and this provides instant indication as to whether a change has been made when problems arise.
3.1 GEORGE 4 Mark 8.12 (PEB,CJP)
Mark 8.12 was finally brought into normal service; delay was entirely due to an elusive bug which led to filestore loss and which it took much effort to convince ourselves was not due to ACL. The effort was good for the soul and uncovered ACL and ICL faults, none of which appeared to explain the loss. Stepping with care, Mark 8 was used at night for a week or two and eventually the plunge was taken.
Reliability has been difficult to judge as hardware problems have been ever present. There have been record periods without breaks and then patches of very bad behaviour. Resilience from bad hardware is still poor. Also GEORGE 4 is still behind GEORGE 3 both in time and standard of validation it seems.
Efficiency under Mark 8 has not been impressive. However there is no evidence that this is due to Mark 8. MOP activity has increased a lot and clear records have been set for the number of jobs run although the performance for the individual has worsened considerably.
3.2 GEORGE 4 Mark 8.22 (PEB,CJP)
Mark 8.21 was obtained from Reading (while ICL were having problems at Bracknell) but it was shortly followed by 8.22 and this will be our next Mark. Work has begun on this, and it is hoped to be in service some time in May. The move should be easier as confidence in ACL edits is now well established.
Phase 1 (very rudimentary) of the ICL jobwell appears in this; the Oxford jobwell will be merged with this (and gradually phased out in later versions). Automatic restarts of jobs lost in breaks will be one useful feature.
About 2/3 of the chapters have been changed to some degree, but most of the ACL edits can be applied without change. Future Marks will probably change very much less and a change of Marks is hoped to become a much more automated procedure in the future.
3.3 GEORGE development
The short period that Mark 8 has been in has seen the provision of a useful crop of enhancements which have been festering on the sidelines for 6 months or so. There are still a fair number of changes pending but few, if any, major ACL enhancements and stability can be seen about 1 year off.
CONVERT command (PEB): implemented to provide an elementary desk calculator on MOP terminals. This involves a new chapter and it will be easy to implement special permanent and temporary facilities in it. Normally this will not be for general use but will enable measurement or control to be achieved.
LISTFILE paging (PEB): for fast VDUs is now available and proving useful. The framing features are also present and should aid the use of Tektronix displays. With these features, 100 char/sec terminals are far more useful.
Urgency system (CJP): changes in preparation for the new urgency system (see below and TN 107).
Trivial changes have been made to LI command and some journal messages (CJP).
3.4 New budgeting system (CJP,RJW)
The increase in system loading, and particularly the activity of certain discretionary users, have made necessary a far more rigorous budgeting system involving the introduction of 3 urgencies (TN 98, 107). Changes have involved:
- GEORGE, to check the user's urgency request against his time left.
- HLS: substantial amendments to allow for urgencies H and Z and to take more account of guaranteed status.
- Accounting system: to account use at the new urgencies.
(1), and (2) are complete but not fully tested. (3) is partially implemented in that time used at all urgencies is totalled up. The second phase of (3) will produce separate totals of, use at each urgency. The scheme is awaiting user allocations of time at each urgency by Resource Management Branch.
An accounting and database working party has been set up in order to decide on matters of accounting and budgeting etc, which affect more than one group. The two meetings so far have been very productive in making decisions about the new budgeting scheme.
3.5 Accounting and Budgeting (RJW)
The newly structured period accounting system described at the end of last quarter was installed early this quarter and has been running successfully since.
A program to produce a sorted list of each user's SPACEMT budget has been produced. This should be very useful to the tape librarian who previously had to run a job for each user for whom the information was required.
The bit-dropping in Dictionary update records, mentioned in the last two QPR's took a dramatic turn. The bug appeared to have gone in Mark 8 but after a week it was found that complete records were disappearing instead. A bug report has been submitted to lCL.
3.6 TASK (GWR)
The realisation that the stock of TASK manuals was almost exhausted speeded up the revision of the Manual. Despite the favourable comments on the current version it was decided that as so many changes have been made, a complete rewrite of sections 1 and 2 was required and this is now almost complete. The opportunity was also taken to review the specification of some TASK parameters and a rather more logical implementation was devised which has resulted in making the LINK parameter almost redundant. In the light of user experience new facilities are to be added to the binary program peripheral system allowing users to create workfiles and DA files. The removal of the automatic listing of files sent to filestore has required the introduction of a LIST parameter. All these facilities should appear during the coming quarter.
The introduction of the FORTRAN block read compiler was finally successful and no problems have occurred since. Minor ICL mends have been added to other compilers as required. In order to keep track of the latter the compiler setting-up system was modified to incorporate a sum check on each compiler so that it is possible to check if all modifications have been done to both main and reserve compilers. It will also ease maintenance of a TASK system remote from ACL.
3.7 NUTS (DAD,GWR)
This quarter has been largely concerned with the implementation of a system of intelligent printing of magnetic tapes in standard ICL format based on the subfile structure. The system will print in record format data blocks which are in COPYOUT/FORMATTED FORTRAN or UNFORHATTED FORTRAN format.
The FORMAT parameter has been enhanced to provide rudimentary facilities for the user to define the format of the data blocks on the tape. In the initial version the subfile to be printed is specified by name and optionally by a generation number. If the generation number is specified as -0 or +0 then the subfile of highest generation number will be printed.
A more restricted system for DA files in standard formats is also being implemented using the XMED disc editor. This will only list named subfiles.
It is hoped that these enhancements will be available by the end of April.
The facilities of the Kernel have also been enhanced by the addition of a STEER facility similar to that of TASK which greatly aids testing of the system. The feature allowing multiple activation of NUTS in one call of the macro has also been improved and if it proves sufficiently reliable may be released. This has a similar format to that of TASK and may be used in the macro and in STEER files. Other new parameters are DY to change directories (cf TASK) and facilities allowing NUTS to claim Q and S-trusted status for reading and writing exofiles for Operations. Also for Operations, NUTS will now run from the operators console and display back to the operators the results of the call.
Difficulties were experienced when trying to copy certain subfiles into exofiles (see TN 105) which has meant that this facility in COPYIN has limitations. The changes to the code outlined in TN 1O5 have been made, but testing is not yet complete, and the new version will not be released until the changes detailed above are also complete.
The ICL utilities COPYIN 13.nd COPYOUT were updated at the beginning of March with new versions written in PLASYD. Some bugs in this version have since come to light and have been corrected. The changes to the exofile handling system detailed in TN 105 will be incorporated in COPYIN as soon as they have been thoroughly tested in the NUTS version. Detailed documentation of COPYIN/COPYOUT for internal use by the Advisory Office is currently being prepared. An Internal User Note documenting new facilities in these macros will be prepared as soon as possible.
Several enhancements to COPYIN/COPYOUT are currently in preparation. With the current version of COPYOUT it is not possible to COPYOUT two or more files to subfiles with identical names if a subfile with that name is already present on the tape. Copying subfiles with common subfile names into filestore requires at present knowledge of the internal working of COPYIN. Coding is currently in preparation to allow users to COPYOUT several subfiles with the same name, and the generation number of the subfile will be incremented accordingly. COPYIN will be modified to permit a subfile generation number to be specified, and provision will be made for the generation number to be specified as +0 or -0.
Consideration has been given to defining a default generation number - 0 for COPYIN, but since this would involve a large overhead in terms of tape transfers, the idea has been shelved for the present.
Code has been produced and is currently being tested to enable a subfile to be split over two reels of magnetic tape so that NUTS can COPYIN and OUT an entire EDS 60 pack.
3.8 Geronimo (JDT)
The Geronimo program has been adjusted to run correctly under George 4 Mk 8, and the LIST facility has been enhanced to display ALLCHAR files in upper and lower case. The documentation file has been rewritten to provide a full operator specification of the system.
3.9 Microfiche (PEB)
A scheme for the production of microfiche is well advanced and should be available about the time that the FR80 is brought into service. The system will enable output to fiche instead of lineprinter to be achieved with minimal changes to the program. It will be possible for the operator to direct output destined for the lineprinter to be directed to fiche. There will be some restriction in the initial version which will require GEORGE changes to solve.
Journal utilities have been incorporated into a single macro JNUTS, and a technical note produced (CJP, SRP).
A utility is being written (SRP) to analyse and list all jobs which might degrade the normal running of the machine, eg large quota, large number of pageturns.
Statistics on TASK over 3 months have been produced (SRP) (TN 104).
3.11 Dump Tape Editor (SRP)
This has been used to produce small filestores for operator use and for system development.
3.12 Database editor (AI-B)
Most of the time in this quarter has been spent extracting information from a copy of the accounts database for the study of its layout.
It is hoped that FORTRAN utilities will be produced by TREEMETA Edit-compiler in the near future.
3.13 Filestore statistics (CJP)
A further technical note analysing filestore and exofile use over the last year has been produced.
3.14 Large user seminar (CJP)
A Seminar for large users of the 1906A has been arranged on 17th April. The subjects were determined by the results of a questionnaire. Speakers will be CJP, AJHW and PEB.
4. THE 4080 PROJECT (PEB)
The front end processor project is continuing at high pressure. The Kernel and HASP code show signs of completion. EPSS effort has been minimal (due to lack of the duplex board) but it may well be possible to allocate more effort in the next quarter. As suggested the 7903 emulator code is proving large and difficult and further effort may be needed here. The hardware is well behind due to a month's management course by Neil Parker. In addition further rewiring has been undertaken due to poor RL workmanship. The interface has worked during January and there is little doubt that there are no technical problems.
The disc, store and duplex communication interface were delivered to just catch the 1974 financial year.
EPSS liaison group meetings have been attended with interest.
High level protocols have been defined by some groups and they do not seem implementable on the 1906A. It will be interesting to see how well they are adopted. It is likely that only MOP facilities will be provided from ACL on EPSS during the first year.
Discussions with RL are reaching some conclusions on a network interface to the 360/195. This should be defined by mid year and implemented by Christmas. This is later than hoped but the definition is superior to the Edinburgh scheme.
Code is now reaching a conclusion for allowing line testing via the 4080. The Sigma 2 (which in any case now appears more or less permanently broken) can now be disposed of. The problems of maintenance are becoming onerous and no good use has yet appeared.
4.1 4080 Operating Systems (DCT)
No work has been performed on the core operating system COS, except modifying the current version to work with a larger (96k bytes) core store.
Following the delivery of the disc at the end of March, it had been hoped to produce quickly a version of the disc operating system DOS with the required attributes to run on our 4080. This end has so far been thwarted by hardware unreliability. The generation of DOS systems is considerably easier and quicker than for COS, due to the availability of the disc filing system and a primitive macro facility.
4.2 7903 DCP EMULATOR (DCT)
Much effort has gone into detailed design work for this part of the GEC 4080 front end project, and a large number of flow charts have been produced. This stage, which involves making all the remaining design decisions, is now almost complete, and coding from these flow charts can con1ffience in a week or so.
A number of the GEORGE changes required have been completed, in particular the changes to the GEORGE property system. A satisfactory way for the 4080 to issue operation commands via its special MOP job is still required however.
4.3 HASP CHAT (JDT)
The proposal to analyse messages from HASP work stations within the Kernel has been found to be unworkable, and a new process has been designed. This process will analyse textual messages from HASP Consoles and teletypes, and interface them with the Kernel.
4.4 4080 HASP (MDF)
Work is progressing well for the process which deals with the buffer assembly for RJE's. Code has been written for the line handler interface, the signing on mechanism, and for the translation of data between the internal standard for the KERNEL interface and the HASP journal for the line. Limited testing has been done, and the implementation of the code for the KERNEL interface will enable more comprehensive testing of the process.
4.5 4080 Kernel (JDT)
The code for handling file connection and disconnection and data transfer has been written, tested and flowcharted. The assumption that the message interface was adequately defined turned out to be incorrect, so little progress has been made on the LOGON and LOGOFF sections.
4.6 1906A - 4080 Interface (NMP)
As a result of tests carried out in January, parts of the logic have been redesigned and considerably simplified, thus making the possibility of fitting the interface onto one board in the 4080 cabinet seem more likely.
However during the course of changing the logic more bad connections were found in the back-plane wiring which may have contributed to some of the problems we had been having with the termination of transfers.
It was decided that the back-plane should be rewired and this has now been completed.
Tests are about to start again and it is hoped that reliable data flow should be achieved during this series.
No work was carried out on the interface in February due to courses.
4.7 Magnetic cartridge equipment for the 4080 (NMP)
Two decks have been purchased from Data Recording Instruments which use the 3M's DC300A cartridge. Malcolm Harman, a sandwich, student from Plymouth Polytechnic is currently working on formatting electronics and an interface for the decks. They will probably be made to look like paper tape equipment to the 4080.
4.8 Contention unit
This has now been built and tested by David Ashby. It is designed to allow up to 8 channels to contend for up to 4 ports.
A new rack has been installed in the communications area to accommodate the CU and its associated modems. This is being wired in at the present moment.
5. GEC 2050 - 7020 EMULATOR
5.1 EBCDIC Card Input
To allow for EBCDIC code card input to the 1906A via 2050's a version of the Emulator was produced to perform the required character conversion. One of the difficulties is that some 1900 characters do not occur in the EBCDIC character set and so unused character and. multi-punching had to be resorted to.
5.2 Paper Tape input
Two problems have occurred in this area. The first, from SUSSEX, found that the whole system crashed if the first character read in to a buffer had the wrong parity. As there is no paper tape readers on the ACL 2050, the error had to be debugged by telephone to Sussex.
The second problem occurred at ROE, whose paper tapes for input to the 1906A contained no parity - ie the parity column was always punched. The Emulator had to be modified to translate the data to the correct parity and this was done successfully.
ACTIVITIES - Communications and Systems Branch
PAPERS - Internal
- 1906A INTERNAL USER NOTICES
- 100 Macro ALGOL68, S R Perkins
- 101 Macro LSTR, S R Perkins
- 102 GEORGE COMMAND CONVERT (Mark 8), P E Bryant
- 103 Paging and Framing on MOP Terminals, P E Bryant
- 1906A TECHNICAL NOTICES
- 97 FR80 Output, P E Bryant
- 98 Proposed Changes to Budgeting and Scheduling, C J Pavelin
- 99 ALGOL68 SYSTEM JOURNAL Utilities, C J Pavelin
- 100 Tape Library Growth, P E Bryant
- 101 1906A Executive Investigation, P E Bryant
- 102 GEORGE 8.20 POSTBOX, P E Bryant
- 103 Filestore and Exofile Characteristics (4), C J Pavelin
- 104 TASK Statistics in System Journal, S R Perkins
- 105 Use of COPYIN/COPYOUT with Exofiles, D A Duce
- 106 New 1906A Accounting Programs, R J Walters
- 107 Changes to GEORGE and Scheduler, C J Pavelin
- FRONT END PROCESSOR PAPERS
- 44 4080 Project: Desirable GEORGE changes, D C Toll
- 45 DOS Process Shells, D C Toll
- 46 Decisions taken at a meeting on 27 February 1975, J D Thewlis
- 47 Meeting of the GEC 4080 Users Group on 20 March 1975, J D Thewlis
VISITS AND COURSES
- 3-14 March
- Advanced Course on Compiler Construction at the Technical University of Munich, S R Perkins
- 27/28 January
- SRC Visit to IBM Hursley, P E Bryant
- 3, 10 and 17 march
- 3 seminars on the 4080 front end processor project - talks given by - P E Bryant, M D Fowler, J D Thewlis, N M Parker, D C Toll
MEETINGS AND LECTURES ATTENDED
- 9 January
- ICLCUA GEORGE 3 User Group General Meeting, C J Pavelin
- EPSS Liaison Group, P E Bryant
- 16 January
- ACTP Software Committee, P E Bryant
- 20 January, 24 February
- Atlas/Daresbury/Rutherford Network Collaboration, P E Bryant
- 25 February
- Networks Panel, P E Bryant
- EPSS Liaison Group, M D Fowler
- 26 February
- ACTP DAP Meeting at ICL, P E Bryant
- 27 February
- ACTP Software Committee, P E Bryant
- 4 March
- UIJPC Meeting, Birmingham, P E Bryant and C J Pavelin
- 7 March
- EPSS Liaison Group Study Group 3, P E Bryant
- 13 March
- Discussions with British Steel on 4080, P E Bryant
- 21 March
- ICLCUA GEORGE 3 User Group, Performance Activity Group, G W Robinson
- 24 March
- GEC 4080 Users Group, J D Thewlis
- 25 March
- Discussions with Mr King and Mr Crossley, Mullard Research Lab, on 3-shift working, P E Bryant
6. FR80 HARDWARE
6.1 Acceptance (RET,PMN,RB)
At the start of the quarter, the delivery date of the FR80 was moved back to 10 March. The main reason for this was that the hardcopy camera was not going to be as late in delivery from the manufacturers as was first thought. The original plans had been to make the factory acceptance without the hardcopy camera. Therefore, this was not really a slippage but more a decision by III that the camera would arrive in time for them to do their own checks prior to acceptance. Delaying the acceptance tests ensured that the hardcopy camera could be used.
RET and PMN went to Los Angeles on 15 February with the aim of starting the acceptance tests about a week later. However a number of hardware and software problems arose which significantly delayed the start. There was some indication that III had not had sufficient time to set up the machine properly with the tight schedule required by the Laboratory. The situation was exacerbated by the Chief Engineer being away during this period.
The major problem appeared to be a misalignment of images on all the cameras. This was finally tracked down to the collar holding the tube in place being incorrect and allowing the tube to move. This was a direct consequence of III changing to the Litton tube which was a slightly different shape from the Ferranti one.
The acceptance tests did not begin until 5 March and were completed by 10 March with some reservations. One or two pieces of software required modification. The ability to produce 64 evenly-spaced intensity levels was not achieved (only 63 were possible). The clear filter in the colour wheel was a straw colour instead of being transparent. Abutting on 35mm and hardcopy was only just acceptable. It was quite clear that most of the remaining problems could be sorted out at the local acceptance once the machine arrived at the Laboratory.
The FR80 arrived on site on 19 March, no damage having occurred in transit. The only problem encountered during the unloading was that the film processor would not go into the lift and had to be man-handled up the two flights of stairs.
The installation of the FR80 on site has taken longer than expected and the local acceptance tests have still not taken place. With measuring equipment being resident in the Rutherford Laboratory and unaccessible outside normal working hours, the number of trial set-up runs per day has been quite small. The image size on the 35mm camera had changed significantly during transit. Also, the characteristics of the new film processor had to be ascertained before any intensity tests could be done.
Most of the problems seem to be resolved apart from the abutting, on which work is still going on. Two areas where problems were expected have proved reasonably straightforward. The light output of the new tube is sufficiently greater than that of the old one for good quality hardcopy to be achieved at a single hit. Also, reasonable colour output, even at the red end of the spectrum, has been achieved with one hit.
6.2 Operation (RB)
Most of the quarter was spent in preparing for the arrival of the FR80 and 5010 film processor. Providing power supplies was relatively simple. However, the water supply to the 5010 film processor did present some problems. At one stage it seemed likely there would be an insufficient head of water to drive the processor. The water outlet is in such a position that the processor has had to be raised by putting wood blocks under it. Finally, due to the late delivery of a calorifier, it is likely that showers will be cold in A5 whenever the processor is running!
The SD4020 tape deck and some furniture had to be moved to make room for the FR80. It was found that the 1906A console could not be moved as far as was first thought. Consequently, some modification of the intended position of the FR80 had to be made.
The colour film processor was installed in the Rutherford Photographic Section but only lasted for about two weeks before it had to be replaced due to cracks in the plastic and problems with the drive motor.
The operators for the FR80 were selected during the quarter and training was started during the February shutdown. This included a visit to Leeds to see the COMP80 and 5010 film processor owned by the Post Office. A short photographic course was given by RCGW. A training course will be arranged for the operators once the Acceptance Tests have been completed.
DVR and AWB have agreed to act as back-up operators in emergencies.
7. FR80 SOFTHARE
7.1 SMOG (AHF,PAD)
The SMOG package has now been implemented for the FR80. All facilities except file saving have been implemented for the Tektronix and the line drawing routines for the Hewlett Packard pen plotter have been implemented. There have been many changes to the specification of the package made. during the last quarter. These have usually occurred as a result of more information about the FR80 becoming available. The most significant changes have been in the use of text strings and the way in which they can be concatenated. These changes have necessitated several rewrites of the package. A macro to generate the SMOG library has been written, and the macro :SMOG.SMOG is used to run the package. After some initial problems which were exacerbated by the lack of documentation available, NULLIB has been used successfully to generate the library for SMOG.
The specification for SMOG has been continually updated and several versions have appeared during the quarter. It is hoped that the latest specification will remain firm. A first draft of the user manual for SMOG has been written and is being corrected in the light of comments received about it.
SMOG sends FR80 output to either the spooling system or the user's magnetic tape. Line drawing, both relative and absolute, text strings, file saving and format control routines are all written for the FR80. Currently testing has been done through JMR's VIEW program and by listing the FR80 commands since the FR80 is not yet available. Output for the Tektronix is similar to that of SPROGS but text strings can be concatenated as on the FR80. The advance frame messages now use *+*+ in order to take advantage of the new ALL option to LISTFILE.
Output for the Hewlett Packard is restricted to line drawing at the moment. Again the advance frame and colour messages now use *+*+. Text for the Hewlett Packard will be implemented in the next quarter, as will file saving for the output devices other than the FR80 and the lineprinter graphical option.
A manual for SMOG is currently being written.
7.2 SPOOL (JRG)
The purpose of the 1906A spooling system is to gather several small and medium graphics jobs onto one tape for the FR80, via a standard disc file. The graphic packages route the user's FR80 orders to the disc according to a specified layout and a read-out program accumulates them on tape in a form suitable for LOADGO on the FR80.
A similar system was used for the SD4020 but the layout for the FR80 is a complete redesign owing to important differences between the 2 devices.
The specification of the disc layout has undergone several minor changes since being produced in December.
The coding required to route user's graphical information to the disc has been included in the graphics packages (FRAH, RET, AHF).
The read-out program has been written and tested. This can now be operated from the central console via a macro called FR80. It will extract all jobs for a specified FR80 camera and batch them onto one tape.
A program to view frames of a spooled job on the Tektronix has been written and tested.
An enquiry program for the spool has been written and tested. This enables an operator to discover (via the FR80 macro) how many jobs are in the spool disc file and how many for each or a specified camera.
Some utilities are being written at the moment which will allow the FR80 operators to keep track of the tapes used by the spool and also those tapes generated by the microfiche listing facility.
In the coming quarter, changes will be made to the disc file specification and the read-out program to incorporate the accounts data with each job. The read-out program will be extended to write 9-track tapes if required (since the FR80 has both a 7-track and a 9-track deck). Improvements will be made to the way the read-out program deals with failures of tape, disc and user jobs. A utility will be provided to delay the removal of a job so that it can be viewed on the Tektronix by AWB's viewing program.
7.3 Tektronix VIEW (AWB)
This utility has been provided to allow users to examine graphical output destined for the FR80 while it is still in the 1906A spool. The user may run a job which converts specified frames into graphic files which can be displayed on the Tektronix. It is possible to erase the job in the spool if the output is incorrect.
7.4 GROATS (FRAH)
The GROATS package has been converted to generate output for the FR80. This can either be sent to the spool or a 7-track tape. The only work still to be done is the addition of accounting records at the front and back of each job.
7.5 SPROGS (RET, AHF)
Version 3 was released to users during this quarter together with the documentation. So far, this seems to have caused few problems, except for the general problem of size and the amount of lower data used. Since this version deals with more output devices, it is basically larger than the previous version. It is possible to cut down on this size, however, by making use of routines designed to remove unwanted sections of the program The default program description was changed to make it compatible with that provided by TASK.
Attention has now been turned to providing an FR80 version of SPROGS. The FR80 code will replace the SD4020 code, thus making the FR80 available as device 1, 2 or 32. Initially, both versions of SPROGS will be made available, and dummy SD4020 routines will be left in the FR80 version.
A version of SPROGS, using 7 or 9-track tape, or spool, had been written before RET went to USA. This was in a semi-tested state, and made use of the user-supplied device testing facility. Work is now in progress to convert the code to replace the SD4020, and to incorporate all the changes that have been proposed since that date. As SPROGS will also contain all the SMOG routines, the SMOG changes must also be made.
A version of SPROGS using lineprinter graphical and Tektronix (via MUGWUMP) device only, was implemented on 360/195, but proved very large. This was in part due to the high overheads associated with small subroutines. Attention has now turned to the implementation of SMOG, and work will resume on SPROGS at a later date.
7.6 SCSIM (DVR,RET)
The SCFOR simulator, SCSIM, has been written and tested. It uses SMOG to generate output for the FRSO via either magnetic tape or the spool. A User Note indicating the changes necessary for existing programs has been written and will be issued prior to the FR80 service starting. Once SMOG is working on the 360/195, SCSIM will be implemented there.
7.7 CAMP and CAMPER (DVR)
A version of CAMP which produces output on the Tektronix via SMOG exists. Work on CAMP and CAMPER has stopped until effort is available to complete the task.
7.8 SPROGS on the 360/195 (MFC)
The version of SPROGS taken over from AJHW has been implemented for Tektronix 4010 and lineprinter graphical output. SPROGS library files required the writing of low level code:
- to simulate #ELASTIC on the 1906A (achieved using GETMAIN and FREEMAIN macros on the 360/195;
- to implement buffered direct access disc I/O without the severe constraints imposed by the extended FORTRAN direct access system.
Most problems encountered were due directly or indirectly to the difference in word length for reals and integers between the two machines, the remedies including:
- Replacement of calls to AMOD by a mixed real/integer equivalent.
- Imposing a restriction on text string file names of 6 packed alphabetic or three EBCDIC characters.
Work was begun on implementing SPROGS fonts and apart from small differences between the EBCDIC and ICL character sets, a simple software font is available.
The SPROGS effort has now been suspended until work on SMOG is complete. It should then be possible to take a new version of SPROGS (that drives the FR80) from the 1906A to the 360/.195 and recompile the whole system, incorporating the 360 specific modifications and software produced. for the previous version at the same time.
A major drawback of SPROGS on the 360/195 is the amount of core it requires (arising from numerous inter-related external references between a large number of routines). It appears that a minimal subset of routines requires about 200K(bytes) and typical test cases need around 250-300K. Much core could be saved by rewriting the smaller routines in ASSEMBLER (thereby avoiding the FORTRAN overheads of trace tables, prologues and epilogues): this would be a straightforward, though somewhat time-consuming, job. The re-coded routines would have to be made as invulnerable as possible to future changes in the SPROGS system.
7.9 SMOG on the 360/195 (AWB , MFC)
An FR80 handler and order generator has been written using FORTRAN I/O to generate 7- or 9-track tape and may be used as a basis to SMOG or in stand-alone mode. Test tapes (including one using a crude simulator for SCFOR options 1, 2 and 8) have been viewed on the PDP15 using JMR's program, though without FR80 LOADGO headers. The desirability or otherwise of using IBM tapes with non-standard headers and trailers between data sets is being considered.
A version of SMOG has been successfully compiled on the 360/195, though, as for SPROGS, considerable savings in core can be achieved by re-writing shorter routines in ASSEMBLER (about 50 routines have already been converted with a saving of about 30%). As the present 1906A SMOG package is still undergoing changes the very latest version is not yet on the 360/195.
A collection of miscellaneous routines for bit and character handling, location of jobname, account, ID, etc is being assembled for use with both SMOG and SPROGS.
A VIEW program is in the process of being tested that will enable users to send information in FR80 data format to MUGWUMP for viewing on any of the graphical devices supported by RL.
Work on the FR80 spool is being carried out by C D Osland and M O'Connell at RL. Since the 360/195 has been undergoing major modifications, full implementation of HASP spooling of graphical output has had to wait until the coming quarter.
7.10 FR80 LOADGO (JMR)
ACL intends to use the III LOADGO system on the FR80. This allows several plotting jobs to appear on a single tape, each prefaced by a pair of header records. These header records specify the name of the monitor program which is to interpret the data, and also supply a set of monitor commands which are to be obeyed by the chosen monitor before it begins processing the data for the job. A specification has been issued of the interface to be used at ACL between the graphics packages and the LOADGO system.
7.11 FR80 Leader Frame (DVR, AWB, JMR)
Graphics jobs plotted on the FR80 will output a standard identification frame at the front and back of each job. This frame will contain both job-dependent and job-independent information. The job-independent part is to be stored within the FR80 as a permanent picture. Actually a pair of permanent pictures have been defined, differing only in size, since no single size of picture could be found which was appropriate to all forms of output. (Strangely, the FR80 lacks a facility for scaling pictures.) A third permanent picture has been defined which contains a standard resolution pattern. The attempt to construct these pictures revealed some problems with the graphics packages available at ACL, and in particular led to a redefinition of some aspects of SMOG. It is still unclear whether or not permanent pictures can be stored within the FR80 in such a way that they persist from one invocation of the FR80 Displayer to another.
7.12 FR80 VIEW (JMR)
A PDP-15 program has been written to allow the viewing of FR80 tapes. This program will take a magnetic tape containing graphical data written in FR80 Standard Data Format and display the results on the VT04 refreshed display. The program has proved itself to be a useful debugging aid during the process of modifying the ACL graphics packages to generate FR80 code. In its present form, however, it is a tool more useful to the implementors of such packages than to their users. It is hoped that a more elegant version of the program will be provided in the future in order to remedy this defect.
7.13 FR80 ACCOUNTING (JMR,RWW)
An automatic accounting system is to be constructed for the FR80. This will work as follows. All graphical jobs run on the FR80 will be recorded and followed by a logging job. These jobs will supply statistics concerning the graphical job sandwiched between them to a special logging program. The logging program will record these statistics, and also some information which it gathers for itself, in a reserved log file on the FRSO disc. A facility will be provided to enable the manual entry of information into the log. Periodically, the log file will be dumped onto magnetic tapes which will be regularly processed by the 1906A. This processing will use the logged data to compute the amount of FR80 system resources consumed by each job, and so provide the required accounting services.
At present the structure of the system has been designed in outline and a specification of its interface with the graphics packages has been issued. Some understanding of the III software has been achieved, prior to the production of a detailed strategy for the construction of this facility.
7.14 FR80 Software (WDS)
Most of the quarter has been spent in the tangled undergrowth of FR80 documentation. A disassembler for FR80 binary files has been debugged. The program currently outputs instruction mnemonics, but not symbols. Investigation into the possibility of FR80 program development on the PDP15 suggests that this is a long-range project. Patches to the FR80 Displayer to include larger software character drawing and a permanent Â£ character are being studied.
7.15 FR80 Colour Output (PAD)
The colour problem exacerbated by the absence of any useful texts prompted the writing of a Discussion Paper: The Colour Problem and The FR80. Feedback so far suggests that there is still much confusion about Colours when one argues in detail.
A suggestion regarding a method of setting up a colour service on the FR80 was accepted. Accordingly, preparations have been made, including the acquisition of certain apparatus, so that as soon as the FR80 is accepted, an intensive period of hardware (eg filter) and software optimisation will take place - the object being to provide users with as comprehensive and comprehensible a colour service as is economically feasible.
8.1 PDP15 (WDS,RWW)
PDP15 down-time this quarter has been due largely to the addition of the LA36 DECwriter and high speed Tektronix interface. The DECwriter outputs very legible hardcopy at 30 characters/see with both upper and lower case. The Tektronix 4010 now runs at a blinding speed of 9600 baud! WDS is gradually taking over the management of the PDP15 from JRG and PB.
With the advent of the FR80, it became quite clear that the addressing mechanism of both the FR80 and PDP15 were not clearly understood. A new addressing mode was discovered on the PDP15. A number of tests were done which resulted in RWW producing a comprehensive definition of the addressing.
8.2 PIGS (WDS)
Version 2 of PIGS (PDPI5 Interactive Graphics System) was updated to run under DOS V3/RSX. The new version, PIGS V3, is being sent to Latrobe University, Australia, for possible installation on another PDP15.
8.3 Tektronix 4010 and PDP15 (JMR)
Owing to the extreme and intolerable unreliability of the VT04 display, a variant of the FR80 VIEW program was constructed to drive the Tektronix 4010 storage tube, rather than the VT04. While experimenting with the 4010, it was found that it occasionally generated catastrophic interrupts within the PDP15. This was found to be due to the fact that the 4010 concerned had a particular strappable option (namely the full page break) enabled. The device was dismantled and this option disabled. The interrupts have disappeared and the fault is considered cured. Disabling the full page break has had the further benefit of curing a fault which sometimes occurred when the 4010 concerned was connected to the 1906A.
It seems likely that each of the Tektronix 4010's used by ACL may have a different combination of strappable options enabled. (For example, one returns 7 bytes to a certain GIN mode command, while another returns only 5.) Since many of these options affect the operational characteristics of the device, it is felt that ACL should study the options available and standardize on one combination only.
8.4 PDP15 - ]906A Interface (JRG)
The hardware fault which appeared at the end of the last quarter and which causes a block transfer to be suspended has not been cured in spite of much work by lCL and DEC. One of the problems is that it has become intermittent (failure rate ranges from 1 in 1000 to 1 in 10000 blocks) after being hard initially. However, in this state it is possible for the PDP15 to detect the condition and successfully restart the block.
8.5 Projection Room (AHF, PMN)
A plan for a bench in the projection room was finally agreed upon and has now been installed. A list of audio/visual equipment to be housed in the projection room has been drawn up and the equipment, where available, has been installed there. The new Bell and Howell projector has been slightly modified to allow film sound tracks to be played through the Colloquium amplifier/loudspeaker system.
The PAG equipment has still not arrived although one unit is promised for early April. The film winders for the editing bench have been completed but have shown a need for more powerful motors which are on order.
A Kodak Carousel projector is now available as well as a bulk eraser. The latter will cope with computer tapes in addition to audio tapes.
ACTIVITIES - FR80 Project Branch
PAPERS - Internal
- FR80 DISCUSSION PAPERS
- 7 Proposal for Camera Schedule, R Brandwood
- 8 SD4020 Tape Analysis, D Ralphs
- 9 Viewing of Jobs on FR80 Spool, A W Burraston
- 10 Proposals concerning the operation of the FR80/1906A spooling system, J R Gallop
- 11 FR80 Spool Tapes, F R A Hopgood
- Colour, P A Dewar
- FR80 OPERATOR NOTICES
- 1 Outline of Duties, R Brandwood
- 2 Colour Processing at Rutherford Laboratory, R Brandwood
- 3 Safety Precautions in Film Processing, R Brandwood
- FR80 PROJECT NOTES
- 5 Notes on meeting held on 15 January, F R A Hopgood
- 6 Notes on meeting held on 29 January, F R A Hopgood
- 7 Meeting on definition of SMOG, A H Francis
- 8 Notes of meeting and current position at 12 February, F R A Hopgood
- 9 Future Work Programme, F R A Hopgood
- 10 Notes of meeting on 3 March and record of current position, F R A Hopgood
- 11 Notes of meeting on 14 March and record of current position, F R A Hopgood
- 12 Notes of meeting on Accounts Program held on 21 March, W D Shaw
- FR80 TECHNICAL PAPERS
- 2 LOADGO Header Records, J M Rushby
- 3 Larger Software Characters for the FR80, W D Shaw
- FR80 USER NOTES
- 1 FR80 Microfilm Recorder, F R A Hopgood
- PDP15 USER NOTES
- 5 DOS Version 3A/New Version of PDP15 FORTRAN IV, J R Gallop
- 6 Bugs in F4V44 Compiler, W D Shaw
- 6 Addressing Mechanisms, R W Witty
- SPROGS PAPER
- 51 Regions (Default Settings), D Ralphs
- GRAPHICS NOTE
- 3 Standardisation of data files (DMAC), D Ralphs
COURSES AND VISITS
- January 27-28
- Photographic Course at ACL, R Brandwood and Operators
- February 11-12
- Central Induction Course, Appleton Lab, J M Rushby
- February 12
- Visit to Leeds to see Post Office FR80, R Brandwood, D Ralphs + Operators
- February 15 to March 11
- Visit to III, Los Angeles, for Acceptance Tests on FR80, R E Thomas, P M Nelson
- March 19-21
- SRC Management Course Stratford-upon-Avon, R E Thomas
- March 24-26
- Session Chairman, CAL 75 Conference, Oxford Polytechnic, R E Thomas
- January 14-March 18
- Series of 10 weekly lectures on Translator Writing Systems and Compiling Techniques to students at BruneI Univ, F R A Hopgood
- January 21
- Lecture on Computer Graphics at Reading University, R E Thomas
- February 6
- Illustrated talk on Computer Animation to Systems Research Group, Computing Science Dept, Newcastle University, J M Rushby
- March 4
- Talk on FR80 Project to Resource Management Branch, ACL, D Ralphs(with R Brandwood and J M Rushby)
9. GROUP TALKS
- January 8
- Control of Parallel Processors, C J Pavelin
- January 22
- 370/168 and VS2 Operating System, G W Robinson
- February 5
- EPSS, P E Bryant
- February 19
- Advanced Course on Computer Architecture, J D Thewlis
- March 19
- Oxford Medical Computer Centre, T Gover (Oxford Reg Hosp Board)
10. EXTERNAL PUBLICATION
The 1906A Reference Manual (14 authors)
Volumes I and II now distributed.
Volume III expected by end April.