Re: GNU Extension Language Plans

[Michael R. Blair <ziggy@martigny.ai.mit.edu>]

> I found your note fascinating.  I had no idea that RMS's language is
> based so much on a particular implementation.

You included the GNU posting in your own message but apparently did
not read it carefully. It says so just below the list of changes he
wants to make to Scheme to make it more elisp-like and less
Scheme-like.

> You also made it seem RMS has an outdated view of the state of Scheme.

I don't think I made it seem this way. I simply stated outright that
this is the case. You apparently did not read my post very carefully
either.

>  I ask you the [sic] send
> more information about the GNU Extension Plans to the Scheme list so
> that we can offer intelligent and productive commentary on this
> effort.

You already have available to you the same details of the plan that I
have.  I good first step for you, then, would be to go back and read
it more carefully

The only thing I claimed to have access to, beyond that note, was some
personal insight into Stallman's decision making process, having
overheard him discuss the Scheme shell (scsh) and Scheme 48 with Olin
Shivers one or two days before he sent his GEL plans to
comp.lang.scheme (Scheme Digest). After over an hour of seemingly
technical discussion about the relative technical merits of the
various alternatives, Stallman's overriding concern became that the
license agreements the other implementations offered would not allow
him to impose the GNU public license, which tries to make itself
contagious in any piece of software with GNU software inside. He did
not particularly care exactly what the licenses were for the other
stuff, only that they were not the GNU license.

To address the details of his plan in an intelligent and productive
manner, I believe Stallman's list of desired changes to Scheme, for
example, is a good starting point for unravelling his mis-impressions
of Scheme. He says he wants dynamic binding as an extention but makes
no mention of FLUID-LET, for example. He also seems to call for
property lists but makes no mention of symbol hashtables or RECORDS or
DEFINE-STRUCTURE.

I think if you look very carefully at the various language design
issues he raises, you will see that he has not followed the Scheme
language design debates for nearly a decade. Overall, I believe he has
not bothered to browse either the Scheme Digest archives or the RRRS
archives on altdorf.ai.mit.edu. Perhaps he is unaware of their
existence. Perhaps you would like to tell him about them so he can do
his homework too.

In short, I think Stallman has, for years now, isolated himself from
the Scheme community, even though many of us are just a few doors down
the hall, but now he wants to come down from the mountain with his
rough-cut tablets and tell us how he wants Scheme to be. I personally
am rather fond of the highly polished calf we have fashioned out of
gold and I am reticent to just simply give it up because this oddly
behaved bearded gentlemen says that's the way he wants it. Personally,
I don't find his list of gratuitous changes to be an altogether earth
shattering breakthrough in language design principles.

I'm sure you will all appreciate the irony of that last paragraph.

 -ziggy