Text preview for : 19781031_Minutes_October_30.pdf part of xerox 19781031 Minutes October 30 xerox sdd memos_1978 19781031_Minutes_October_30.pdf



Back to : 19781031_Minutes_October_ | Home

Inter-Office Memorandum

To Debugger Working Group Date October 31, 1978


From Barbara Koalkin Location Palo Alto


Subject Minutes - October 30 Organization SDD/SD

litO! SDD AaCijlVES
XEROX I have read and und~ (..'
Pages_________TQ~
Filed on: [IRIS] D5 >DWG10-30.BRAVO
Reviewer Date_ _ .
I of Pages Ref .2SSPi) -~as-
Present at today's meeting: Ayers, Johnsson, Karlton, Koalkin, Satterthwaite, and Wallace.
The following is a summary of suggestions discussed at today's meeting:

* Progress has been made by Smokey and Phil towards creating a model of the
debugger/WindEx using the Tools environment. They expect to have a workable model by
mid-week so we can do some experiments.

* For the time being, type-in goes to the window containing the cursor (as in the Tools
Environment).

* We all agreed that variables should be interpreted in the stack window in the context of
whatever is "bulleted" (Le., marked). We will have to do some experiments to see whether echoing of
the input characters will be done in a subwindow of the stack window or in an interpreter window.
However, the Debug.Typescript equivalent window will be capable of doing "contextless"
interpreting (Le., arithmetic).

* To change the context: select one of the context keywordS and:
* push down the menu button. A menu of the choices will appear. Letting up the
button over one of the choices (or outside the menu to indicate no change) will make that
the context. We still need to look into how we will display a large list of information (as in 50
modules in a configuration) and indent to show the structure (for nested configurations).
* OR type in the new context terminated by CR. Smokey can provide a mechanism for this
sort of "field fill-in".

* There is no "THE current context" (as refered to in the previous memo); however, each window
has a context that is used for symbol-lookup (and marked in some way to indicate tilis).

* The initial window configuration will contain (at least): one stack window (initialized with the
information associated with the process that is currently running), one source file window, and a
debugger communication window (Le.. Debug.Answers. or a Debug.Typescript equivalent).

* We need a mechanism for telling alllhe windows (a notification) that "an bets are off' (i.e.,
a Ilew program has been loaded. some files have been deleted) so that it knows when its cache
information might be invalid.

* Smokey pointed out that the PNR nolificr mechanism of the Tools environment can
implement the same flow of control scheme as the current debugger (with respect to unwinding
the stack on proceed. procedure calls, etc.),
Minutes