I asked for explicit scenarios that you would like implemented--that
is, exactly what keystrokes and button pushes you would perform to
achieve the desired tasks. I want suggestions, not complaints.
Time taken to start TLC. I assume most of the cost is due to spawning
a new JVM on every invocation, and most JVMs are very slow to start.
Simon Zambrovski tried to run TLC inside the Toolbox's JVM, but
failed. He was unable to write the appropriate code to re-initialize
everything so a single instance of TLC could be run more than once.
My memory says that he re-initialized everything he could find to
re-initialize, but it still didn't work.
Lack of a pretty-printer for complex expressions
TLC pretty-prints all values that it outputs. If you think you can do
a better job, it should be easy to find and replace the existing
pretty-printer. In fact, the current one was written by Mark Tuttle
because he didn't like the way Yuan's code was printing values.
However, Mark is a pretty bright guy and I suspect that you wouldn't
do much better. I think you'd find that every change you made to make
it work better on some output will make it do worse on other output.
You might be happier with a structured viewer like the one for values
of variables in the trace explorer.