@chapter Other Software Issues

@section Toolbox Calls

A proposed K-to-Mac major command, CALL-ROUTINE, 
would load a `MacR' resource by name, and optionally pass shared-memory compatible arguments.
These routines may be normal Macintosh routines.
They could be available to the user from the K/Mac application menu.

Upon invocation, the K/Mac could open a specified set of such `MacR' resources.
This could be used to implement convenient ``patches'' to the application.

As another example, the LISP Machine's @l(initialize-timebase) return value
could be provided by the K/Mac via a `MacR' CALL-ROUTINE.

You can trivially create a Mac interface routine to load and open an
additional resource file.  Then a KFSL file can, when loading, call this
routine to open itself as a resource file, and have its own Mac-side
interface routines as well as Lisp side, all in one file.

This supersedes an earlier proposal using general purpose entry
vectors (simple #'s) to gain access to subroutines with mainline
calls.


@section Interlocks

The proposed interlocking of commands and buffers between the K and Macintosh
is done with a flag word or byte.  
The values of the proposed flags are as follows:

@itemize @bullet
@item
0 = NULL-STATE == thing is available,
action is not required
@item
1 = BUSY-STATE == action is underway,
waiting
@item
-1 = DONE-STATE == action is completed
@end(itemize)

For some flags, values other than 1, 0, and -1 may be used to indicate
error or status codes.

It must be clearly established in each context
which of the two processors is exclusively
responsible for manipulating a flag. 

[Has a better scheme been proposed? RWK: YES, in MTL. We need a single
interlock scheme for buffers and command blocks!

@section Areas for Future Discussion

The following are areas for research and future projects:

@itemize @bullet
@item "KFSL" MacFiletype

Possible extensions to K/Mac application.
For example, clicking on an executable .KFSL file
could boot the K/Mac with initializations to
run the selected program.

@item Integrating KFASL

Integrating QFASL and KFASL formats -- within ZMacs,
8-bit bytes, with the compiler.
@itemize -
@item Integrate Jim's 8-bit QFASL change
@item Integrate KFASL with RG's "target processor"
in Lambda compiler/ZMacs code
@end(itemize)

[Deferred until future meeting among Peter, Jim, Keith, and Bob.] 

@item How much work will be required to check out
and, if needed, clean up canonical types and
generic pathnames?

@item Conditionalizations

There will be a wide variety of configurations (per processor, software, bus links).
We need to support  read- and run-time conditionalizations for each case.
This must wait until the migration and integration plans have been written.
[GSI needs to generate a separate specification.]

@item Cambridge / Montreal communications

We would like to ensure that the Macintosh group can
reliably use UUCP mail for intersite communications.

@item Lambda microcode

Sources have been provided to Montreal to facilitate first-pass interface designs.

@item Lambda/Macintosh configuration

Montreal is working with a facility for Lambda/Macintosh
communications that allows for simulation of the interface.
We would like to have the means to do this as well.
This will facilitate testing procedures before the K has been completed.

Test areas: Replace kbd, mouse with Mac's [paging should be a last
resort].
Incrementally design and test interface and communications area,
moving gradually away from microcode.
Use Lambda programs as simulators.

@item Soft deletion issues

As a future enhancement, we might consider a file command
to view contents of trash folders (deleted files).  
See section @ref[section-softdelete] for proposed scheme.

@item An inter-filesystem consistency-maintenance mechanism
(counters, ticks, semaphores, locks?)

@item Documentation, validation, and test code

If possible, we would like to review any test software
used for validation of the K/Mac software developed in Montreal.
Integration with the LISP sources will be done by GSI,
and some (re-)testing will be required.
Also, design (theory of operation) documentation for the Macintosh application code
will be very helpful.

@item Keyboard redesign

The keyboard re-layout project, including the
character set mapping, is still pending.

@item BYTE layout (LDB)

Suggestion: If the new, K-specific byte-specifier formats 
can be changed to be compatible with the Lambda,
some extensive LISP software modifications could be avoided.

@item Apple licenses and registration issues

Registering filetypes, application names, etc. with 
Apple?  What is required?

@item Nubus issues

Find out whether Macintosh NuBus behavior affects the ability
of other processors to become the bus master.
If so, what impact does this have?

Issue: Are K-to-Mac interrupts ``guaranteed''? ...Mac-to-K interrupts?

Question: How is data provided to the Mac on an interrupt?

@end(itemize)

@textbox
@center Montreal notes:
Login handled from the K side.
@end textbox