[Aldor-l] [Axiom-developer] Re: exports and constants
Gabriel Dos Reis
gdr at integrable-solutions.net
Thu Jul 27 14:37:01 EDT 2006
"Page, Bill" <Bill.Page at drdc-rddc.gc.ca> writes:
| Gaby,
|
| On Thursday, July 27, 2006 1:40 PM you wrote:
|
| > ...
| > Bill Page writes:
| >
| > | It is very very easy for the compiler to generate self-
| > | identifying type constants either all the time or as a
| > | compile-time option. This can even been done in a simple
| > | pre-compilation step if necessary.
| >
| > Either we have a very different view of what "Reflection"
| > is, or we have a very different experience in the use and
| > implementations of that suff :-)
|
| That is quite possible. Care to give a couple of simple
| examples we can continue to discuss?
Well, first it is you make the claims that it is "very very easy" to
implement; I would say the burden of proof is upon you :-)
However, my starting example is getting right the Figure 1 of the
Template-Haskell paper. Can you provide a proof-of-concept as to why
it is very very easy to do?
| > I don't find Template Haskell implementation "very very
| > very easy". Here is an overview of Template Haskell
| >
| > http://research.microsoft.com/~simonpj/Papers/meta-haskell/index.htm
| > http://www.haskell.org/th/#papers
| >
|
| I have read some of these papers (you've referred to them
| before) and I am inclined to agree. I have not actually tried
| to code in Template Haskell yet, but do not find this extension
| of Haskell particularly "comfortable" yet. It's true isn't it
| that some of the goals of meta-programming go beyond what one
| would normally associate with language features that support
| reflect?
full-fledged meta-programming yes -- and Template Haskell is not there
yet. But by "reflection" I imply the bare minimum currently provided
by Template Haskell for example.
| Perhaps the difference between Template Haskell and what seems
| easy in Aldor (at least with respect to self-identification and
| reflection) has to do with significant differences in the
| underlying type system.
I'm not so sure. Aldor's type system appears to me to be far more
complicated and intricate than that of Haskell. Both share a good
deal of functional core, though.
| Aldor is set apart from Haskell for
| historical reasons. Perhaps Aldor has more in common with ML?
|
Hmm, in what aspects?
-- Gaby
More information about the Aldor-l
mailing list