At 1:11 PM -0800 on 2/10/00, Alain Farmer wrote:
>> Eric: ... but script errors do happen,
>> and it is really nice to be able to
>> watch global variables.
>
>Alain: Yes, one of HC's most endearing features is the
>ease with which one can debug one's scripts. The
>variable watcher, and the step-by-step debugger too.
I know... certainly have missed it in Bison. It's one of the things
being put into NuParser, though.
>Alain: Both of these should be external and programmed
>in low-level code, just as it is now in HyperCard,
>such that the debugging stuff remains efficient and
>transparent.
Message watcher could be (sort of) done by catching every message and
passing at the home stack. PITA, though.
>instead of
>re-inventing the whole thing in FreeScript.
Why not do the whole thing in FreeScript?
>Alain: It depends on the XCMD. If your XCMD was made
>in order to hide/protect your code and/or make
>repetitive actions faster, then it is likely that
>these XCMDs will continue to run properly on all
>platforms. The problem arises when your XCMD
>capitalizes on platform-specific features.
XCMD's are compiled code and by definition must rely on
platform-specific features, such as the microprocessor they are running
on. x86 code won't run on a PPC or vice-versa. Also, all the API's are
different.
Even if you use 100% strait ISO C++, and avoid anything that could
possibly break (hahaha), it'd still need recompiling for every platform.
>Alain: In HyperCard, the Script-Editor, the
>Variable-Watcher, and the Message-Watcher are external
>resources, coded at the same (low-)level as HyperCard
>is. What's more, you can substitute these resources
>with your own variants of these debugging facilities.
>Isn't this how we should do it too?
Why? Much easier to code it in FreeScript. Much easier to modify it
that way, too.
