Jump Over Left Menu
Quarterly Progress Report 1 October - 31 December 1974
F R A Hopgood and P E Bryant
16 January 1975
In November, the organisation of the Group changed significantly as the result of a management decision to establish a single Branch to be responsible for all aspects of the running of the FR80. As a result, the Basic Software Group is now split into two separate Branches - the Communications and Systems Branch and the FR80 Project Branch. Dr P E Bryant heads the first (with Doris Hopkins as secretary), while F R A Hopgood heads the second as well as retaining leadership of the Group.
As a result of this re-organisation, Paul Nelson, David Ralphs, Tony Burraston, Roland Brandwood, Mike Chiu and John Lewis (Â½-time) 'joined the FR80 Project Branch, while Anton Walter returned to the Applications Software Group to work on ALPS.
Wade Shaw and Andy Dennis left the Group during the quarter to take up appointments abroad. Peter Dewar joins on 1 January.
The Group structure now 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
- A H Francis
- J R Gallop
- J W E Lewis (Â½)
- J M Rushby
- P M Nelson
- R Brandwood
- D V Ralphs
- A W Burraston
- F R A Hopgood
It had been hoped that the highlight of the quarter would have been the introduction of Mark 8. However, the effort on Mark 8 has taken a leisurely pace keeping in step with its introduction in the other 1906A sites. A number of attempts have. been made to introduce it but in the interests of maintaining a good service its use was discontinued on the slightest suspicion of faults. The signs are encouraging for its imminent adoption.
The performance of Mark 7 has been maintained but few changes or enhancements have been introduced. This has led to the accumulation of several enhancements pending Mark 8. These enhancements will take at least 6 months to implement.
2.1 GEORGE 4 Mark 8 (PEB, CJP)
Most GEORGE effort has been concerned with Mark 8. All ACL mends have been converted to source EDITS and tested satisfactorily. The ACL version of Oxford's Jobwell has also been converted (there were few problems as this system was composed mainly of new chapters). The Bradford Mark 8 split chapter file has also been incorporated. This involved recompilation of FIXED which presented some minor problems.
All the ACL mends have been numbered and a scheme for administering and keeping track of them has been implemented.
Certain utilities used in GEORGE development have been modified to work in Mark 8. This includes the QUICKMENDS system to list the binaries of, or modify, chapters. This now reads in the chapter table first, since the disc address of chapters can vary. Also included is the program to list all GEORGE messages and the scheme to produce a sorted list of chapter names with other details.
Mark 8 has given little trouble so far. The only known outstanding problem concerns the apparent changed format of binary programs due to the correction of a Mark 7 bug. The correction involves corresponding changes to Mark 7 (to enable regression) and to TASK. The solution has been to remove this correction which can be included at leisure. Interestingly the elusive PAGELOCK fault of Mark 7 has followed us into Mark 8!
The next quarter (hopefully) will see Mark 8 established. Major enhancements scheduled are Upper and Lower case on RJEs, an archival scheme, paging of output to fast VDUs. Several minor enhancements are envisaged such as shortening of the sign-on message, some means of suppressing the sign-on file, a simple desk calculator.
Thoughts are maturing on changes needed to GEORGE to enable the GEC 4080 front end to operate correctly. A scheme to divert output to microfiche is under consideration.
Other sites are also active and ACL hope to implement the Oxford Mail Box Scheme. The Swansea 7903 changes to enable 1900 networking are also to be examined.
Once GEORGE 8.12 is established 8.20 is likely to appear swiftly and certainly before the end of the quarter.
2.2 High Level Scheduler (CJP)
Changes stimulated by the introduction of Jobwell are continuing. Where possible an estimated start time for a job issued from a MOP job is broadcast back to the terminal. This time, updated every 2 minutes, is also included in the file examined by the STATE macro.
Scheduling on 9-track decks has been re-introduced. The operators can now set, using the SCHED macro, the number of 9-track decks available to submitted jobs as distinct from RETRIEVED. Thus if decks are broken or a large number of files are being retrieved, tape jobs can effectively be streamed by setting the number of decks available low.
The number of 9-track tapes required is again given in the STATE macro output and thus on GERONIMO WELL command to enable the policing of requests.
2.3 Performance (CJP)
The number of MOP slots has been raised to 14 and the number of jobs allowed per user raised from 2 to 10. The increase in MOP activity has been the principle feature of the quarter.
Attempts to introduce the BEATTIMER (real time clock) have played havoc with performance figures. Initially it was calibrated incorrectly and subsequently appeared to vary. Recently it was found to be on when thought to be off (human error).
The hardware problems appear to be cured and the timer is being checked against the hardware monitor with benchmarks. It is worth pursuing since in principle it is a more accurate way of measuring performance than hitherto available. Initial indications are that the time used by jobs and GEORGE has been undermeasured and thus executive time has been overmeasured.
2.4 Accounting and Budgeting (RJW)
At the end of this quarter testing is in hand of the last part of the restructured central period accounting software and it is hoped to install the new system early in the next quarter. The system now consists of three programs interfacing via common files. The first extracts accounting information from MASTER DICTIONARY and outputs it in a simple format graphic file. The second analyses the graphic file statistics and produces the various group periodic usage accumulations which are stored in the user information data base. The third resets the amount used part of the user budget records in the dictionary by subtracting the figures from those in the records. This last part is accomplished by organising the input data into structured stacks and uses the new budget extracodes to update the dictionary.
The bit-dropping bug reported last quarter has been traced to a feature in one of the budget extracodes at EMS time. The feature has been reported as a bug.
The new period accounting system will allow easy re-running and fault recovery, normal running in any environment and possible introduction in future of Journal accounting by means of the graphic file now produced in part 1.
The current production accounts schemes have continued without fault during the quarter. Some modifications have been made to the file to SD4020 transcription utility program originally written for the accounts. This is now available for general use via the COMFILE macro. The new version accepts a LEADER parameter which gives a leader and trailer of 100 blank frames. This version is also more efficient by suppressing some unnecessary output.
A list of usernames holding TRUSTED privilege was produced to enable a purge to correct a historically untidy system.
Proposals have been formulated for future budgeting and scheduling strategies for the 1906A. Decisions on this should be made shortly.
2.5 Data Base Editor (AI-B)
An editor for the accounts database is being produced. This takes the form of a translator or compiler which takes a string of editing commands and outputs FORTRAN statements. The translator has been generated using TREE-META. It is hoped in this way the editing language will be easy to modify or extend.
The syntax for the basic editor has been designed and checked by TREE-META. Some test code has already been generated although serious work on the code generation is only just beginning.
The syntax and other documentation is not yet published but can be inspected in the file :NSIN35.DOCUMENT.
2.6 Dump Tape Editing (SRP)
A dump tape editor is now complete and has been used to generate small filestores for the testing of GEORGE Mark 8. It should also be for useful re-organising parts of the filestore.
2.7 System Journal (SRP)
The Journal compression system has been altered. The archiving of the files runs on a similar system to the full Journal and removes the necessity to generate several steering files. It is more secure and will obtain new tapes as required.
2.8 ALGOL68 macro (SRP)
A new macro has been written to enable ALGOL68 jobs to be run in a similar if more limited manner to TASK.
2.9 LSTR (SRP)
A new LSTR listing level is now available (MEDIUM). This provides most of the relevant information in a high level listing but condensed into one line per directory entrant.
2.10 NUTS (DAD, GWR)
The arrival of DAD has more than doubled the effort on this project and major progress has been made this quarter. The file :NEWS.NUTS-DOC documents most of the new facilities.
COPYIN and COPYOUT have both been implemented under NUTS in an enhanced form. COPYIN can read files from a magnetic tape in any previous COPYOUT format as well as basic peripheral files from GEORGE dump tapes. The latter feature should soon extend to DA files.
Facilities have been introduced to COPYOUT exofiles to magnetic tape, and t o COPYIN subfiles into exofiles. If a DA file is copied to a magnetic tape using the ICL macro COPYOUT, the file is assigned with a WRITE or OVERLAY qualifier, which could lead to loss of the file if a GEORGE break occurs whilst the file is thus assigned. Under NUTS such a file is assigned for reading only.
Work is currently in progress to provide facilities for copying multifiles to and from magnetic tape.
The ACL COPYIN/COPYOUT format is in accordance with the latest ICL formats and will therefore be compatible with all other ICL sites.
Efficiency should be improved by the proposed change to block reads for basic files.
A lot of effort has gone. into the facilities for 7-track NONSTANDARD tapes. NUTS can now read and write to any mode supported by the ICL decks. NUTS has overcome an Executive anomaly on reading certain tape modes when no count of the tape block size is given in the PERI reply. It can also read and calculate the size of tape blocks containing parity errors and this means that more information from damaged tapes can be recovered than hitherto possible.
The specification of the mode of a 7-track tape has been enhanced with MODE parameter(s) signifying the actual difference(s) from the Standard mode as an alternative to octal or decimal parameters: For example 200 BPI,EVENPARITY,SHORTGAP. This reduces the complexity and possible mistakes in this area. When reading from 7-track tapes, NUTS will check that the tape is in fact readable in the mode given (or implied by default) and, if not, will attempt to read it in all the modes available.
When using any tape, NUTS writes out details of its speed, parity and interblock gap, and the appropriate MODE, to the monitoring file.
It is thus possible to ascertain the mode of any tape that comes to ACL if it is readable on our tape decks.
The PRINTing of any magnetic tapes and disc files has been enhanced by the addition of three more formats, HEXADECIMAL, PLAN and TEXT. PLAN causes the information on the tape or disc to be interpreted as PLAN instructions (cf GEORGE PRINT command for Core dumps) and can be qualified by EBM or DBM to print the appropriate branch mode. EBM is the default. TEXT if used as the only FORMAT specifier prints the information as text strings and can be qualified by a number to give that number of characters per line in the printout.
An additional FORMAT specifier is ABS which corrects the inversion done by the ICL 7-track decks on reading or writing tapes, and allows tapes from other machines (including the FR80) to be printed out in a more intelligible form.
NONSTANDARD tapes may be qualified by the POS parameter which is used to give their rack number in the ACL tape store to the operators before the tape is requested.
Other improvements to NUTS include the LOG facility (as itemised for TASK) and the addition of CDC code to the card conversion facilities.
Rough documentation (:NEWS.UTILITY-DOC) is available for users to write their own utilities to run under NUTS and DAD and SRP have done this successfully.
Plans for the future include intelligent printing for magnetic tapes (cf MTPRINT) and disc files based on their subfile structure. The code conversion system is also going to be improved to indicate where in a file any unconvertable characters are, and also to extend the feature to magnetic tapes.
The ICL utilities COPYIN and COPYOUT have been rewritten in PLASYD, and have been extended to include the additional facilities and modifications embodied in the NUTS versions.
The new utilities are run by the macros NEWCOPYIN and NEWCOPYOUT under :MACROS.
2.11 TASK (GWR)
With the final and welcome demise of the FORTRAN and ALGOL macros, effort has been concentrated on improving and enhancing TASK.
In the light of experience the compiler system has been redesigned so that the introduction of new compilers is easier, thus the introduction of TASK to other sites needs less modification to the code. Duplicate copies of the compilers are now held in exofile and filestore files, and TASK obtains its compilers from the exofile unless a read failure occurs when it switches to the filestore duplicate with no ill effects to the job being processed. A similar exofile/filestore file pair hold the compilers currently under test (obtained by the TASK parameter TEST). When these are satisfactory either a new version of TASK can be put up which regards these as the main compilers or the current version can be switched from the operators console to use these compilers. In the event of problems with the new compilers, the old versions can be quickly reinstated from the operators console. The compilers in use (and the state of the other settings outlined below) are indicated in the language code of the current version of TASK and this is used by the compiler setting-up system to ascertain which are the current test compilers so avoiding any possibility of corruption of the production system.
Other enhancements, also switchable from the operators console, concern the FORTRAN compilation system. When a file is not specified for the semicompiled output TASK can use either a system exofile or files tore file, scratch space or a work file depending on how it is set up. It is thus possible to evaluate the various .methods of running FORTRAN compilations and hence to optimise what is a fair proportion of the workload in terms of jobs run.
For example, using system exofiles limits the number of compilations and consolidations running in parallel to the number of exofiles available hence allowing non compilation (and usually more CPU bound) jobs to proceed. This has been tried successfully with two exofiles for some time but problems arise in the fact that there is no means of ascertaining which exofiles are being used so a random selection has to be used. Thus during long compilations several jobs can be queuing for one exofile whilst the other is free. Experiments have been suspended for the moment until more time can be found for a proper evaluation of the system and Mark 8 is fully implemented.
So far, the introduction of the new FORTRAN compiler which uses block reads to read in the source program has been unsuccessful. This should give an improvement in compilation efficiency as it reduces disc traffic. An earlier introduction of a similar feature into the PLASYD compiler went very smoothly. The FORTRAN system has proved rather troublesome, but introduction is imminent.
More successful was the introduction of the ACL version of the XPCH consolidation which no-one seems to have noticed.
A very useful new feature is the LOG parameter. This causes TASK to append one or two records for each TASK on to a file in the users main directory. This file is then automatically listed and erased by the STATE macro. Thus the use of the STATE macro will give the user a complete picture of his jobs since its last use, giving either the jobs queuing to be run or concise message about the result of any TASK jobs. Several other systems (eg NUTS and ALGOL68) will have the LOG facility.
The use of unusual binary formats by several ICL overlayed utilities was solved by modifying the TASK RUN System to recognise such binaries and to take the appropriate action to initialise the overlay areas.
The main section of the TASK macro has been rewritten to make the monltoring file even easier to understand especially in the region of running a binary program. Other improvements include the enhancement (suggested by a user) of the #DA parameter to online a zero length scratch file if no filename was given and the reduction in the number of files used by TASK (especially when using LINK) to improve efficiency.
Work on the new version of the TASK manual is proceeding slowly. Thanks are due to those users who made suggestions for improvements to the manual and TASK. Thanks also for the very complimentary tone of the majority of replies.
2.12 JUGGERNAUT (GWR)
Modifications received from ICL were incorporated to cure several failings in the system but a general restore showed up further bugs. These have now been cured and the system is now satisfactory.
2.13 1906A Reference Manual (FRAH, JBC, DAH)
An edited version of the three volumes of the 1906A Reference Manual was produced in this quarter. DAH and Denise Barrett completed the typing in a remarkably short time. The first volume was passed round various members of the laboratory for commend and error-finding. JBC coordinated most of the changes that were needed. The first volume was sent to the printers in December and should be available before Easter. Work continues on the other two volumes and these should be sent to the printers in the next quarter. This has been a time consuming project which has taken up a considerable amount of the Group's resources during the last eight months.
The installation of further lines and terminals is now largely in the hands of Operations Branch. However, considerable discussions have been held on the future direction of development. Pressure for further capacity and differing sorts of access is heavy. It may be unwise to develop along ad hoc lines to solve short term problems since this could provide a strain on support effort and present problems in the future. In particular schemes to build a contention unit to multiplex ports into the 360/195 and 1100 computers are being examined. Discussions with RL have resulted in agreement in principles to allow this development. Unfortunately, some of the back-to-back modem tests required have been held up pending the delivery of spare modems. A pair to be lent for this purpose was cancelled due to a mix up. The new type of modems to be purchased will allow a single MOP channel. It has been a great drawback that RJE users have had little MOP access since internal users find it so useful.
The modifications to allow 1200 bit/sec terminals is now proving its worth with the new batch of terminals which can run at this speed. It is hoped that the upper and lower case can also be exploited by these terminals.
3.1 GEC 4080 (MDF, DCT, JDT, PEB)
The major communications effort has been directed to this project. Negotiations with RL on the preferred method of access to the 360/195 continues. The options are access using HASP protocols and concentrator code, HASP using contention code over several lines or access via the proposed network entry to the 360. Regretfully the network entry is not yet well defined and the concentrator method is to be preferred.
3.2 4080 Operating System COS, (DCT)
Development of COS has continued up to the point at which shortage of core renders any further extensions difficult. This development will continue with the disc based operating system DOS which will be delivered shortly. The extra core (32K) will also help.
The command processor has been enhanced to provide extra commands and facilities. In particular a facility has been provided whereby a program can issue commands at run-time.
3.3 7903 DCP Emulation (DCT)
This is the major piece of code in the system. The work is progressing well. It is hoped to have a large proportion of the remaining code complete by the end of the coming quarter. Initial testing with GEORGE can begin in the summer. There are still several problems to resolve in the GEORGE 4080 interface. Some GEORGE changes in the property system will be required.
3.4 KERNEL Code (JDT)
The code provides the intelligence of the system and the switching function. Most of the quarter has been spent investigating the role to be played by the KERNEL process. This is now thought to be sufficiently well understood for definitions of the interfaces with other processes and of the internal data structures to be written. A fair amount of code has been written under the assumption that these definitions will not suffer more than minor amendments. The main framework of the KERNEL appears to be workable.
A program has been written to assist the testing of the KERNEL process and of other processes in the 4080. It is controlled from the console and provides facilities for composing messages, sending them to the process under test and examining replies. It has been used with the prototype KERNEL to simulate an RJE handler and a mainframe handler. It has also been used with the 7903 emulator to simulate the KERNEL.
3.5 4080 HASP (MDF)
During this quarter a line handler for Binary Synchronous HASP protocol has been written and comprehensively tested. This handler will drive both remote lines and 1ine(s) to the 360/195.
The handler deals with control records ACN, NAN and ENQ and data records which may consist of several buffers. For output Binary Synchronous Protocol is added and removed on input. It has been found necessary to set up a receive immediately after a transmit has terminated in readiness for the reply. This is occasioned by the fact that the software paths to higher levels of protocol required before a reply is requested can be too long and cause loss of data.
The CRC check calculation by software has been implemented and functions satisfactorily. A hardware unit will shortly be delivered since the CRC check is expensive in software.
The flow controlling process which interfaces between the handler and the KERNEL has been designed and partly written. The design aims are such that the same code can be used for all lines, that is, lines to remote workstations and to the 195 without the concentrator facility.
Currently code is being developed and tested for the decoding of the RJEs sign-on card which delivers the station number and configuration details to the 4080.
3.6 EPSS (PEB)
Preliminary code is being developed to deal with EPSS protocols. This is very much of an experimental nature as details of EPSS protocols are not yet well understood. The work is at a fairly low level since the network will not be ready for service until September. It is envisaged that MOP entry to the 1906A will be the initial aim since with only one port to EPSS at only 2.4K bit sec. Workstation entry is not viable and is difficult to test. None the less a preliminary design for an RJE station has been written and work may start during the summer when effort becomes available. This will be on the 2050. Liaison with the PO on EPSS has continued and is fruitful.
3.7 Interface to 1906A (NMP)
The Mk 1 interface has reached the stage of passing small blocks of data successfully from the 4080 to the 1906A. Transfers in the opposite direction work so far as mechanising the transfer are concerned but a problem still remains in the termination of the transfer. This should be overcome in the next week or two.
Currently transfers are limited to 13 words (ie 52 24 bit bytes) but it is hoped that this can be increased fairly easily and the interface should be ready to transfer large blocks of data successfully in both directions at the end of January.
A recent modification to improve reliability will enable the interface to be used from the beginning of February for software development and testing. The modification was the rewiring and replacement of certain sockets. This was caused primarily by a broken contact but revealed very poor workmanship on the RL wiring. In fact the work done by RL has been almost entirely redone at ACL. It had been thought the interface would never be reliable but the situation has improved with the modifications. The software development should reveal any amendments that need to be made to the current specification.
Work has started on the Mk 3 interface which is intended to meet the specification and to be installed directly in the 4080 cabinet, hence avoiding the need to use the local external interface board. The Mk 2 which was to be a re-engineered version of Mk 1 may be abandoned as Mk 1 looks like succeeding.
3.8 Contention Unit (NMP)
The design of a unit to allow up to 8 RJEs to contend for up to 4 ports into the 360/195 has been designed. Parts have been obtained and some boards are under construction. It will hopefully be available for February/March.
3.8 Other Hardware
Various devices have been examined. These have been various printers to replace the 4080 lineprinter when the 2050 is shipped to site. Mini magnetic tapes have been looked at as possible I/O devices to allow the removal of the paper tape I/O and for use in various data collection purposes.
PAPERS Communications and Systems Branch
- 1906A INTERNAL USER NOTICES
- 95 A Program to tidy FORTRAN Programs and to produce Flowcharts, ME Claringbold
- 96 GEORGE 4 Mark 8 Trials, P E Bryant
- 97 Oxford Utility Routines available to Internal Users, K M Crennell
- 98 OpScan Device, P E Bryant
- 99 Important Mark 8 Trials, P E Bryant
- 1906A USER NOTE
- 43 Oxford Utility Routines, K M Crennell
- 1906A TECHNICAL NOTICES
- 93 Report of GEORGE 3 User Group Performance ~ Activity Group Meeting in London on 31 October 1974, R J Waters
- 94 ACL Mends in Mark 8, C J Pavelin
- 95 Report of Meeting of GEORGE 3 Business Users Group 21 November 1974, C J Pavelin
- 96 Time Budgeting and Scheduling: possibilities for the future, R J Waters and C J Pavelin
- FRONT END PROCESSOR PAPERS
- 28 Proposal for Line Handler to 360/195 and HASP RJEs - Structure and Interface, M D Fowler
- 29 Forward Look, P E Bryant
- 30 Specification of SIGNON and LOGON Routine, J D Thewlis
- 31 Connections to GEC 4080, P E Bryant
- 32 SIXA: The 7903 Emulator - Basic Structure, D C Toll
- 33 Some notes concerning SIXA - the 7903 Emulator in the GEC 4080, D C Toll
- 34 Use of the KERNEL simulator process, J D Thewlis
- 35 Specification of 1906A - 4080 Hardware Interface, D C Toll
- 36 Proposed Structure of the RJE Driver for GEC 4080, M D Fowler
- 37 Connection to 360/195, P E Bryant
- 38 The Design of a Contention Unit, P E Bryant
- 39 Format of Data Buffers passed between HOST and RJE STATION Processes , D C Toll
- 40 Rutherford/Atlas/Daresbury Network Collaboration Meeting 2 - Minutes, P E Bryant
- 41 HASP Remote Concentrator - Proposed Specification for Interface, P E Bryant
- 42 EPSS Bulk Terminal, P E Bryant
- 43 ACL 4080 Project Requirements and Timescale - A joint review of the RL 360/195 connection to the ACL 4080 Front End Processor, P E Bryant and J W Burren
For the first half of the quarter, work continued on the projects described in previous reports. With the formation of the FR80 Project Branch, it was necessary to concentrate effort on only those projects directly relevant to the installation and provision of a service on the FR80. Consequently, work in some areas has been severely curtailed.
4.2 PDP15 Hardware (JRG)
The hardware fault which prevented any programs being run in core above 32K was eventually fixed and RSX tasks can now use all of core. Two other faults appeared but disappeared before an engineer could see them. The first of these was the PDP15 switching itself off and on at random. This is a fault that has been with us for quite some time and the cures never seem to be permanent. The second fault was that one of the tape decks randomly spaced the tape forwards and backwards of its own accord!
Other minor faults appeared and were cured during the quarter. Those outstanding are one on the BSI and the perpetual keyboard/sparkpen interference problem. This is only likely to be cured by purchasing another tablet. Unfortunately finding a suitable replacement has proved difficult.
Two new items of hardware were ordered during the quarter and should be installed this Spring. The first is an extra line to connect the Tektronix 4010 to the existing LT19 multiplexor. This will enable the 4010 to be used at a speed of 2400 bauds. The second piece of hardware is a 30 chars/see DECwriter which will replace the existing system teletype. This should speed up the interaction with the PDP15 and allow listings to be made at a much faster speed.
4.3 RSX (RET, JRG)
Very little work was done on RSX this quarter. The proper release of RSX PLUS III arrived in October. However, the version of DOS provided still had a bug which corrupted the disc and so no attempt was made to implement this unsupported DOS. A fully supported version has just arrived and was mounted on the PDP15 at the end of the quarter. The version of DOS provided appears to work and the aim will be to use this version as the standard system. The major change between the new version and the one in use is that the FORTRAN compiler generates different code.
Due to FR80 commitments, it is unlikely that any work will be done on RSX this quarter. Consequently, a multi-user system on the PDP15 for users will not be made available in the near future due to lack of manpower.
4.4 BSI (JRG)
Programs to send data both ways across the link were completed this quarter and testing is well advanced. The system program PIP can send and receive data so that files can be transfered to and from the 1906A. Two FORTRAN programs (one in the 1906A and the other in the PDP15) can also communicate with each other although this has not been fully tested. Documentation is being produced but as a low-priority activity.
The BSI continues to be plagued by hardware faults. In October, BSI transfers crashed the 1906A. This was eventually traced to a fault in the 1906A PPU.
Another hardware fault required the co-operation of both DEC and ICL. The stand-alone tests on both machines did not reveal the fault, even though it was virtually solid when the two machines were connected. It took only a morning to sort it out. Contacts were cleaned up on both sides and a suspect board on the PDP15 was changed. Unfortunately, as both were done together, it is not known whether the culprit was the PDP15 board or the contacts. At the end of the quarter, another fault with identical symptoms has appeared. This is still not cured. It is hoped that time will be found during the next quarter to clear it up.
4.5 SPROGS (RET, AHF)
The main effort this quarter has been directed towards issuing the next version of SPROGS, together with documentation, and also to implementing SPROGS on the 360/195. The various changes that have taken place are:
- LPSIZE command allows the basic raster of the lineprinter graphical device to be changed. Thus rough graphs can be drawn at the Teletype without a considerable number of trailing blank lines.
- COMMON blocks have been constructed to provide preset named variables corresponding to the devices. Users can include these COMMON blocks, and hence refer to a device by name.
- An indicator has been provided to allow multiple calls of STSPR without error, and the standard device routines such as TKTRON, SDHARD etc now call STSPR. Thus, for the simple user, only one initialising routine is required.
- A number of internal changes have taken place to speed up the production of lines. In particular, region arrays are accessed directly (rather than by function) and the colour testing has been simplified. One of the internal display routine tables has been removed, since it is no longer required.
- Some changes have been brought about due to the 360/195 implementation. The FORTRAN PAUSE statement has been replaced by a subroutine, and a text array containing the names of display routines has been altered from real A8 to integer A4.
- Log and shear region conversions are now available.
- The GRAPH package has been completed. This enables a user to plot graphs easily. All scaling, axes and graticule drawing, titling etc can be done automatically by the package. Alternatively, the user can supply his own. Multiple graphs can be drawn and the standard background (scales, axes, graticules etc) can be saved in a file so that it only needs to be computed once for a whole set of similar graphs.
A supplement to the existing SPROGS manual has been prepared describing the new facilities and providing an index for the original manual. This should be available next quarter and will be issued when the new version is complete.
A special version of the overlay system for SPROGS was set up to allow PAFEC to use SPROGS. Some assistance was given to Alan Beard who encountered many problems while merging these two large programs.
4.6 ANTICS (JMR)
An attempt was made to gain an understanding of the ANTICS computer animation system constructed by Alan Kitching. The lack of documentation hindered detailed enquiries, but the gross structure of the system is now understood and privately documented. An appreciation of the facilities afforded by the system has been achieved and some limited use made of it. Its author has recently described an introductory exercise in the use of the system under the title A day with ANTICS which gives step by step instructions for the production of a standardised piece of animation. Future plans will include attempting this exercise.
The Contract for the FR80 was finally signed in November. Both before and after this event, a considerable amount of time was spent in meetings and discussions with III concerning both the hardware and software to be provided. RET and PMN will be going to Los Angeles in February to perform the acceptance tests. The estimated delivery date is the middle of March and we are hoping to provide some kind of service in May. Therefore, there is a great deal of work to be done in a short time.
III's decision not to allow us source listings of the FR80 software has, as expected, proved to be a major annoyance. JMR, in particular, has spent a considerable part of the quarter attempting to obtain a consistent definition of the system from a number of inconsistent manuals. One result of this is that some software design decisions will have to be left until after the FR80 has been delivered and tests can be done to see what the software actually does.
The large number of packages available on both the 1906A and the 360/195 has meant that it is impossible to recode all these for the FR80 in the time available and with only limited manpower. Consequently, some attempt has been made to evaluate what the users need in the future as well as obtaining a quantitative assessment of the use being made of packages at the moment.
5.2 Graphics Seminar (PMN, RET, AHF, JRG)
A Graphics Seminar was held for existing and potential users at the end of October. The main aims were to acquaint users with the existing software, to indicate the extra features that will be available on the FR80 and to obtain user's views on omissions from the existing software. About 80 people attended and joined in the discussion sessions with enthusiasm. On the software side, the major criticism seemed to be the lack of a small and fast package for the user requiring simple graphical output. A number of the users with large packages had found difficulty in interfacing to the larger and more sophisticated packages currently available. A second criticism was that the existing manuals need to be supplemented by primers for the new user. As a direct result of these comments, a new package, SMOG, is being implemented on both the 1906A and 360/195. Some work has also been done in producing a primer for SPROGS by DVR. This will include a number of examples showing how to use the various graph-plotting routines and an example film.
Several users and potential users commented on the absence of GINO-F from the set of packages available. The only real superiority this has over the existing packages is the large amount of pressure and publicity being used to make this a UK standard. It is unfortunate that this is taking place with little consultation or consideration. Although inferior to the existing packages, it is likely that ACL will be forced to mount it. Preliminary discussions with the CAD Centre have indicated that they will let ACL have the necessary documentation to produce an interface for the FR80. The aim is to mount GINO-F on the 1906A initially and then transfer it to the 360/195. This is a low priority project and little effort on it is envisaged before April. The situation has been complicated by RL making independent arrangements with the CAD Centre.
A shorter version of the seminar was given at RL to members of C&A division and various HEP graphics users. This has made RL aware of the changes necessary and these will be done jointly by RL and ACL.
5.3 SMOG (FRAH, RET, AHF)
SMOG (System for Mini Output of Graphics) was defined during the quarter. The aim is to provide a small and fast package which will give access to most of the features of the FR80. Routine names are compatible with SPROGS so that users who outgrow the basic package can switch to SPROGS with the minimum amount of inconvenience. SMOG will only allow output to one device at a time and will not provide line-thickening or dotting., As an exercise, the original SMOG definition was implemented using SPROGS routines and showed a 2.5 times gain in speed.
It is hoped that SMOG will be available when a service is provided on the FR80. Users having their own private systems should be able to switch to SMOG with the minimum amount of inconvenience. SMOG will also be used as the low-level interface for packages such as CAMP and CAMPER.
5.4 FR80 Programs (AHF, JMR)
Although III do not provide source listings of their FR80 software, a small set of basic routines is available for incorporation into user programs. A magnetic tape containing these was obtained on the visit to Los Angeles in September. A program has been written for the PDP15 to decode FR80 source tapes and obtain a listing on the 1906A lineprinter. Some insight into the structure of the FR80 software has been obtained by examining these routines in detail.
5.5 SPROGS (RET)
Work has begun on an FR80 device handler for SPROGS. This is being tested using the User Supplied Device feature. A basic handler capable of writing 7 or 9-track tapes is now available. To this will be added a SPOOL feature (once the design is completed) and the necessary initialisation routines. Later, the handler will require extension when SMOG routines are provided in SPROGS. Existing SPROGS programs should be capable of generating FR80 tapes now.
5.6 GROATS (FRAH)
The use of GROATS on the 1906A is not high and, consequently, the aim has been to provide a system that will allow existing programs to generate FR80 output but not to provide additional routines for accessing new facilities available on the FR80. The FR80 version currently produces correct 7-track FR80 tapes. In the next quarter an interface to the SPOOL system will be provided, together with the necessary initialisation routines.
5.7 SCFOR (RET, DVR)
SCFOR is the oldest and most used of the 1906A packages. Like GROATS, it tends to be orientated towards the SD4020. Rather than implement a separate SCFOR package for the FR80, the aim has been to amalgamate it with the other packages.
A set of name-compatible routines has been written using SPROGS. The routine SCFOR, options 1 to 10, have been tested and thus provide the graph-drawing facilities. Basic raster and test routines have been provided but not fully tested. It is hoped to run some existing programs through this system to see how it compares with the SD4020 version of SCFOR. At some stage, it will be possible to speed up the system by replacing SPROGS with SMOG. In this way, it is hoped to persuade users to change from this package to SPROGS or SMOG, each of which will be interfaced to all the ACL graphical devices and utilities.
5.8 SPOOL (JRG)
A new spooling system has been defined for the FR80. By providing suitable facilities for viewing and manipulating spooled output, it is hoped that most of the graphical output will be produced via the SPOOL system. A LOAD and GO system available on the FR80 makes this the most sensible method of running a number of small jobs on the FR80.
A complete definition of the SPOOL system is now available and implementation will occur in the next quarter.
360/195 (MFC, JWEL, RET)
The main work done so far is the implementation of SPROGS on the 360/195. If this is completed satisfactorily. the need for POLYGRAPHICS will be reviewed. Several meetings have taken place with RL staff to consider how best to do the reprogramming for the 360/195.
RL intend to provide routines to convert MUGWUMP files to FR80 format. It is expected that this will be completed by April. As the SCFOR package can be made to generate MUGWUMP rather than SD4020 output. it should be possible for SCFOR to be run in this manner. Once the SMOG version of SCFOR is available on the 1906A. the intention is to transfer this to the 360/195 so that a cleaner interface to the FR80 is provided.
RL are intending to provide a spooling system for graphical output using HASP internal writers. Some code needs to be obtained from Denmark. Once this arrives. it should take about a month to implement the necessary HASP modifications.
5.10 User Support (DVR, AWB)
A considerable part of the time has been spent in familiarisation with the packages available. The main requirement at the moment is to obtain a good idea of the current use of the SD4020 and the likely use of the FR80. This information has been obtained by a number of surveys:
- An informal survey of the large 1906A and 360/195 users to determine who they were and how much use was being made of the SD4020.
- Timing of individual graphics jobs through the system. The main purpose of this was to highlight any bottlenecks. The major delay on the 1906A is probably de-spooling. At times this has been erratic. Frequently. there is considerable delay if only one job has been output to the spool.
- A complete record of two separate days work on the SD4020 was obtained. This indicated that a number of jobs were not using the spooling system even though they were quite small. There was also some indication that increasing the size of the spool area for each job would be advantageous.
- A comparison is being made of the number of FR80 and SD4020 commands produced by the same program. The aim is to get some idea of the likely reduction in output.
- A survey of all 1906A and 360/195 users has been made to get some idea of their graphical needs. Replies are still coming in. An analysis of these will be made during January.
The NCRE system, which uses the SCFOR package, experienced a number of difficulties during December. It was necessary to produce a PLASYD program for editing NCRE's files. Some re-organisation of the NCRE system was also carried out to improve it from both the user and operator's point of view.
5.11 Operations (RB, PMN)
Some time has been spent evaluating the current SD4020 and film processing procedures. RB visited Leeds for a week to get first-hand knowledge of operating the FR80 and the 5010 film processor. A useful checklist of actions to be taken on camera-changing and set up were obtained.
Some site preparation for the FR80 has already been done. The SD4020 tape deck is to be moved to a new position and some 1906A peripherals moved to give more room for the FR80. The aim is to run both the SD4020 and FR80 next to each other for the first few months.
To ensure that sufficient viewers for FR80 microfiche will be available, an order for 10 Kodak Ekta1ite viewers has been placed. These are capable of displaying a full page of either 42X or 48X reduction images.
A colour film processor has been purchased and is sited in RL photographic section. This will enable up to 60 ft of colour film to be processed locally.
Investigations into the suitability of various film stocks and chemicals for the FR80 have been carried out as far as is possible and stocks ordered.
5.12 Future Work
Most of the work for the next quarter has been outlined in the previous section. An FR80 version of the VIEW program on the PDP15 needs to be implemented under RSX. It is likely that this will use PIGS. The form of the FR80 software to be provided by III has been defined. There will be a certain amount of system work initially to get this to run on the FR80 in the form required by ACL.
6.1 Audio Equipment (AHF)
The Laboratory's new audio equipment has now been installed in the Colloquium. The speakers have been mounted in front of the white boards and close to the ceiling. This was necessary to ensure correct stereo separation. The brackets used to mount the speakers were found to be unsatisfactory (one speaker fell off the bracket!). After discussions with the manufacturer, some improved brackets were obtained. These are better but still not completely satisfactory.
The amplifier and record deck are installed in the projection room. Some plans for benches for the projectors to stand on have been drawn up in co-operation with PMN.
PAPERS - FR80 Project Branch
Paper on CAMP and CAMPER published in Computer. Animation, edited by John Halas, F R A Hopgood and D Ralphs
PIGS Manual, W D Shaw
- SPROGS NOTE
- 34 Notes on Graphics Meeting, W D Shaw
- FR80 DISCUSSION PAPERS
- 1 The FR80 Spool System, R E Thomas
- 2 Leader Frames for FR80, R E Thomas
- 3 MOGS (later changed to SMOG), F R A Hopgood
- 4 Some figures on 1906A Spool, A W Burraston
- 5 Visit to Leeds Computer Centre, R Brandwood
- 6 FRESCO Output, F R A Hopgood
- FR80 PROJECT NOTES
- 1 Documentation, F R A Hopgood
- 2 FR80 Branch - meeting on 4.12.74, F R A Hopgood
- 3 FR80 Branch - meeting on 5.12.74, F R A Hopgood
- 4 Proposed Project Timescales and ID frame definition, R E Thomas
- FR80 TECHNICAL PAPERS
- 1 Basic FR80 Tape Commands, F R A Hopgood
Papers on block used on SD4020 tapes and full day SD4020 processing, D Ralphs
Report on Visit to USA and Canada, J R Gallop