The main purpose of the visit was an exchange of views on DEC software/hardware. Edinburgh already have the RSX Graphics Source Code. There is a separate V handler available so we do not need to rewrite our own. They have a bridging system which allows them to switch easily from DOS to RSX and it may be useful to get a copy of this.
They have rewritten the light-pen tracking routine so that control remains in the routine as long as new lightpen hits are received. This means that a much faster response occurs. They also have a double hexagon tracking cross which apparently was in the handler when they received it!
They have generated a number of programs for displaying grey levels of microdensitiometer output using the display. It appears as if these could be useful to our own people.
All parts for the synthesiser are now available apart from a few items that can be obtained from stores. It has been decided that instead of using the plugboard to connect the synthesiser to the PDP15, the connection will be made to a board at the back of the plugboard. It should then be easier to differentiate between additional plugging and that needed for the interface. There should be some flexibility provided in the plugging to the PDP15.
It is hoped that the synthesiser should be ready for action in about six weeks. To some extent, like everything else, it will then be waiting for RSX3.
One problem has arisen when a DRAW is called within a file definition. If the file is not in store then it must be loaded at the end of the filestore. However, this is the point where the file definition is taking place. The possible solutions are:
(1) Insist on a RETRIEVE command before each file definition so that all files appearing in DRAW statements are already in store. The disadvantage is that the user may be requiring text printing in which case a large number of retrieves would be required or the user may be calling library routines.
(2) The size of the item to be retrieved is known. Therefore the item currently being defined can be moved down to make room for it.
(3) It would be possible to define new files at the end of storage and move them to the end of the filestore only when completed. This has the disadvantage of moving every file after definition just for the few cases when retrieves are required.
It was decided that (2) was the correct approach with (1) being allowed as an optional feature to allow the user to work efficiently. It is perhaps worthwhile doing a garbage collect just before doing the retrieve. This might recover sufficient space so that the item need not be moved.
After one week of trouble free running, the faults with the tablet have reappeared. It is in some senses worse now as characters can appear to be received from the keyboard without touching a key. This means that, at times, it is possible to lock the tablet out while the system waits for the remainder of a command to be input from the keyboard.
Several decisions have to be made in the near future:
(1) The current pushbuttons must be replaced by switches together with lights. If these cannot be put in place of the current buttons then a special box will need to be produced. In this case it may be convenient to have lights on the button box and also below the display in the current position.
(2) If the tablet continues to fail then it may be necessary to cut the front off the display in order to keep keyboard and tablet away from each other. It is unclear whether it is a capacitance effect through the operators body or pick up from the spark pen. It is also possible that the tablet is not earthed correctly. It may be worth using the DMAC keyboard near the tablet to see if this is also affected. Also it is possible that using rubber gloves could be a cure if it is not picked up. This may help to isolate which problem we have.
A simple font on a 160 x 160 grid has been developed for identification on the initial frame. This allows 7 characters to be output giving the users name. If 8 characters are required, it may be possible to close them up a bit. They are currently located in the centre with 20 raster units on either side.
