Re: please advise....

[Ginger Booth]

Hi, Nelson,

    Thanks for your response.  And interest in solving the problem for me,
but....  Time's up.  I need to solve the problem for me.  So...real question
is, do you have any advise on how to proceed/what would be easiest/what is
most in line with what Swarm intends to do later, based on the tools available
today?

    On the three possibilities I mentioned:

    1. Tcl script - it's not clear to me that the machinery allows a Tcl
script to drive the binary, only a way for Obj-C to execute Tcl.  Is there
a way to export hooks to Tcl?  Gut feeling:  the syntax is a misery for my
purposes.  (Users = biology grad students/profs, not hackers.)

    2. Dynamically linked Obj-C module - dunno how to do this either.  Nor
is the syntax much better for the target customers.  Most easily extensible...
    
    3. Write a custom interpreter.  The default.  Your suggestion is as good 
as any, if you're fairly sure that's the way you're going to go.
    
    On the possibility you mentioned (a control file is a text format of a
probe):  I can easily see how this would appeal to most.  This makes one's
probes (popup menus) the "real" thing, and the protocol file just a text
image of the "real" popup menus.  Unfortunately, I consider that backwards.
I fully expect popup menus for writing protocols to fall off the bottom of
my list of things to do, possibly forever, but at least until we've done
enough experiments to add enough interesting hooks.  The "real" thing is 
the protocol, and that must be entirely flexible, extensible, readable, 
editable.  Whether Ginger is a weird special case or not is for you to
decide. :)  The rest of the beta gang is free to chip in their opinions. 

    That said, there's no reason on earth I can't use the same syntax and
machinery as everyone else, at the individual variable level.  And if you
relaxed the probe machinery to operate on "list of probes--arbitrary
collection", I could use it.  I got pretty lost on my limited experiments with
probes.  Whatever I thought I wanted them to do, they insisted on protecting
me from doing. ;)  I do indeed want an arbitrarily long list of mismatched
thingies, container and contents to be determined at run time.

    So.  Would you be willing to flesh out what that syntax needs to be and
the code to read a line and set a variable?  Or how to hack together a probe
with value(s) and action?  With popup menu redirected to the bit bucket?

    In other words, I'm really looking for design advise, and maybe a few 
lines of code for a leg up the learning curve.  Me lost.

Thanks,
    Ginger