🚀 go-pugleaf

I've been experimenting with porting a moderately large (500K source,
100K code + 80K resources) application from Lightspeed Pascal to MPW
Pascal. This isn't because of any inherent faults in Lightspeed
Pascal - it's just so that I can generate a test version of
my program that is compiled for use on 68881-equipped Macs.

The port went fairly smoothly, but I was amazed at how long a typical
build took - to rebuild the whole program from scratch took about
45 minutes in MPW; this is a task that Lightspeed Pascal does in 15
minutes, including build time to write the completed application
to disk.

At any rate, I noticed that a considerable amount of compilation
time is taken sucking the source files for units in a USES clause
from disk, and the following occurred to me: Suppose, if after
a successful compilation, the compiler were to write a .Dump
file to disk? The .Dump file would contain the symbol tables
(just the same as dump files created by the $LOAD directive),
but upon looking for this unit in the future, the compiler
would first look for the .Dump file, and if it's found, use
it instead of textually including the unit.

Is there anything wrong with this?

Of course, the compiler's *still* dog-slow... :-)

	-R.