Re: Re: Web development

[Sergey Dolin]

> > > Surely there's nothing to be gained by converting to XML instead of
> > > S-expressions?
> > I don't know what S-expressions are. Are they  have DTD?
> This terminology goes all the way back to 1958 and the Lisp 1.5 system.
Ok. I will read about it someday. You have intrigued me.
 But one question more: I feel "tagged" and "markup'ed" text is very
 closed to lisp source, but how DTD (or meta-language) can be implemented?

 Goal is: describe classes of my apllication in some language (IDL or
 XML for ex) and when parse this description to make a sql-script
 building database scheme (create table, create table... add constraint,
 create index), c-source to make smob's reading from this tables and
 goops classes definition.
   Now i'm using an h-header files "salted" by comments, but this approch
   have some disadvanages. In XML i can describe many attributes (set of
   valid values, relations and soon) in "standard" way.
  "Standard" -- means, what somebody more clever than i am already
  did work, and cleaned out possible "rubs"(***)

> > I dont know how to store "pointer to current continuatin"... i know it
> > is a shame for "real schemer", but i am as i am. 
> This a triviality:
>   (set! cc (call-with-current-continuation identity))
A know this example :(
  Let's go one step further: there is "dispatcher"-closure which
  keeps set of "continuations". On some input it decides which of
  continuation it should call next.

> I do not really know what you two are trying to do and am not
> paying much attention, but I very much doubt that "continuations"
> (in the Scheme sense) have anything to do with making a transaction
> processing system with roll-back. 
Yes. The mess happened (caused by my imaginaion of continaion as chain
of environments).
  Continuations are good for "dispatching" incoming http-queries. But
  for transactional system we must to store "old" values of variables,
  keep list of them and do all we must do with "regullar" bindings.

  The ideal (as i see it now) would be possibility of "creating" bindings in
  paricular environment. So, "commit" will only kill this environment,
  "rollback" -- will sync "current" environment with it, and we can
  easily implement (touched? ...) predicat by checking does some binding
  have same symbol in "environment for storing".

  But this should be entirely transparent, so we have to break standard
  procedures "define", "let" and so on which is not acceptable.

  Anoter way: to limit "transactional objects" by goops objects -- I
  defined metaclass which can make classes with "transactioanal"
  attribute of slot, but can implemenet storing of those slots only for
  "atomar" values.


(***)
 "rub" came from Hamlet's "...To die, to sleep. To sleep...Oh! There's a rub!". 
 I've not found "rub" in my dictionary but am feeling this word means
 some unexpected trouble.

-- 
Sergey
Mutt/1.2.5i*FVWM2*[FreeBSD 4.3-RELEASE]*(486DX2-50/20/730 Digial HiNote cs450)