Re: Languages: The Bigger the Uglier (was: Re: Aliasing in ISO C)

jgm@CS.Cornell.EDU (Gregory Morrisett) wrote:


> especially when there are other tractible possiblities. Don't you
> readily accept the idea of presenting the syntax of a language using
> a BNF grammar (or perhaps even an LALR(k) grammar)?


I readily accept the idea of using a formal grammer, but that's for a
different reason than you might suspect.


> Then why protest a formal definition of the semantics, using, say
> structured operational semantics?


Because I don't know what you are talking about when you use the term
"structured operational semantics". By definition semantics are both
structured and operational..


> Should we go back to describing syntax using English? I think not.
> It's time to move on...


I think that the point for a formal description is being lost here.
Perhaps I have a different perspective, since I speak both German and
English, and grew up listening to my father describe if .. then
.. else statements and the like, but I think that the entire point of
translation to a specification language is to provide a proof of the
completeness of the abstraction of the language.


It's been my experience in many situations that people that don't
agree on something simply haven't worked through all of the details.
By expressing the concepts in a different language than English, the
participants are forced to really abstract the concepts and then the
concepts can be shown to be complete.


The point of a formal specification is that they are consise and
therefore suffer less from the natural human variations that occur.
It is when these formal specifications can be used in a mechanical
tranlation (ie. a compiler) and are general and consise enough for
general human use as well that we start to find a language that is
useful for "Computing".


In the context of being human, I find it rather easy to "mechanize"
certain types of functions. The point of the mechanization is that it
allows me to complete certain tasks without expending any of the
limited amount of attention that I have available. BNF as an example
feels a very mechanical and therefore allows me the freedom to not
focus on the semantics and therefore allows me to focus on the
information.


But as a child, I was guite a geek and spent a lot of time reading
mathamatical texts and the like. I actually used to carry around
books with lots of formal mathematics. BNF actually feels a lot like
math to me, and given that there aren't many people that can
comfortably read math, (just like a second language), I have to assume
that the same issues are at work. Most people just don't have the
language skills in the first language, let alone the ability in a
second one. When I write software it often feels that I'm just
translating from one language to another.


I think that in the final analysis, it all comes down to a question of
elegance, ie, completeness of the concept, in a consise, simple way,
but then all great design is a matter of elegance, no matter what sort
of design that you are talking about, compiler, language, a piece of
software, or the design of a car.


Wilbur
--

Return to the comp.compilers page.
Search the comp.compilers archives again.