Syntax & Sematics [was: Re: [Enough already] Re: Proposal3:

Andrew C. Greenberg werdna at gate.net
Wed Jun 7 01:00:37 UTC 2000


>  > It seems to me quite naive to think that this, as it introduces
>>  several new degrees of freedom in expressiveness in a langauge, will
>>  not with that introduce several orders of magnitude in complexity of
>>  development and maintenance.
>
>I do not agree on this. Notice that the real living languages, English,
>Swedish, etc, have always had this freedom. Communication has always
>been possible. Somehow we know how to do this, to some extent. For programming
>languages the bijections keeps it together, they dont come for free, 
>they put limits on
>the degrees of freedom.

Comparing the purposes of programming languages with that of natural 
languages is problematic.  Natural language is historically 
ineffective at describing things with precision.  (Indeed, my former 
profession as an attorney would be wholly unnecessary if this were 
not the case).

To me, the code is only part of the issue (which is why I don't use 
pretty printers much) -- the presentation of it is EVERYTHING.  I 
agonize over formatting my production code to try to make it clear 
and maintainable.  I do not see how that can be preserved over 
compilation and decompilation, even when coding straight squeak.

>  > How are we to even DISCUSS code in a diverse open forum, when our
>>  code itself must first be qualified by the parameters of our operator
>>  precedence settings?
>
>We are doing that now, in English, any common language will do.

We aren't discussing a piece of code, whether it works (or works 
well), and trying to diagnose what is needed to repair it (or make it 
better).   We are discussing programming more generally, which is an 
entirely different thing.

>BTW this
>is a real problem look at comp.lang.X, how many discussions do you think
>have been done unknowing of each other because there is no connection.
>Or rather, I would say, people believe that there are no connections
>between the myriad of languages.

True, but this much CAN be said.  If you take a bit of code from a 
listServ, copy it into your system and execute it -- that code will 
very likely have the same semantics.  Not so if I am discussing text 
copied from my programming window.

	4 + 5 * 6

depends upon my system browser selections?  Great start for 
confusion, regardless of whether a "cannonical" version is used.


>  > I remind the community that this is an open source project, and hence
>>  a development effort undertaken by a large number of people.  Even
>>  providing features such as parameterized precedence for binary
>>  operators seems to me a recipe for failure.
>
>We dont agree, I have stated my position as best I could, and I will not
>continue this forever.

You don't agree that this is an open source project undertaken by a 
large number of people?  Surely that is not controversial.

On the merits of whether having identical code in a workspace 
executed differently depending upon personal settings impeding 
communications, I suppose we will have to agree to disagree.
-- 
Andrew C. Greenberg		acg at netwolves.com
V.P. Eng., R&D, 		813.885.2779 (office)
Netwolves Corporation		813.885.2380 (facsimile)
www.netwolves.com





More information about the Squeak-dev mailing list