[plt-scheme] Interacting w/ MzScheme

From: Robby Findler (robby at cs.uchicago.edu)
Date: Fri Dec 10 21:27:23 EST 2004

At Fri, 10 Dec 2004 17:17:44 -0500, Don Blaheta wrote:
>   For list-related administrative tasks:
>   http://list.cs.brown.edu/mailman/listinfo/plt-scheme
> 
> Quoth Robert Bruce Findler:
> > I don't think that I suggested that this was false. But yes. The
> > external format isn't documented. I have tried to explain this before,
> > but I think that I'm just not doing a good job.
> > 
> > The way to manipulate these files is _not_ to write a tool that reads
> > the binary format, but instead use the existing tool that reads the
> > binary format (mred). MrEd gives you back an object you can manipulate
> > directly to do whatever you want with the "file". It's the same
> > information, but directly accessible. It's like writing a parser, to do
> > whatever, but without actually having to write the parser. Code Re-use!
> > Does it make any more sense this time?
> 
> I understood your argument the first time.  It sounds just like all the
> other application developers that say "you can access the file all you
> want, as long as you do it with our tools."  When Microsoft says stuff
> like that, a lot of us argue that it's a bad thing.  Why should it be
> okay when PLT says it?

Please. This is completely different. MS says "we don't tell you the
format and we're gonig to change it so that any tools you might build
will fail in the next release." PLT says "The format is completely
open. In fact, we give you a program that reads it into an _object_ and
lets you manipulate the object." If you want to take that object, write
a (trivial) program to write the data back into a file so you can
re-parse it, go ahead. But why?

> With respect to the bug in ProfJ, that part's obviously not really your
> fault.  And I haven't looked at this in a couple months, so I don't know
> the deep inner workings of the system, so apologies if I've misanalysed:
> it would be nice if what you had to do to register your snip class with
> drscheme was the same as what you needed to register it with mred, to
> prevent bugs like this one. 

It is. As I said, profj just didn't finish the integration. This is not
a slight on profj; it's getting lots of other things right and that's
not really a high priority.

I feel like I'm beating a dead horse with it's glue as I'm jumping over
a muddy pond, so I'm just going to have to let this lie now. Back to
real life.

Robby



Posted on the users mailing list.