axiom-developer
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Axiom-developer] Spad and its object model


From: Stephen Wilson
Subject: Re: [Axiom-developer] Spad and its object model
Date: 12 Jun 2007 10:18:18 -0400
User-agent: Gnus/5.09 (Gnus v5.9.0) Emacs/21.4

Gabriel Dos Reis <address@hidden> writes:
[...]
> That is precisel where I disagree.  
> The current representation forces uses of integer as index.
> And I'm back to my question:  when you index the vtable with value 2,
> what is the meaning of value 2?  You have not answer that question so far;
> except saying you are opposed to any change.

I think that Tim's suggestion of getting a document together detailing
the layout of domain vectors would be a good idea.  I know what most
of the elements of a domain vector represent, and I assume you do as
well.  I hope we can cooperate to produce a document which details
what we have come to understand.

> | Global change to a hash would certainly cost us in execution time.
> 
> Why does that cost more than what you have proposed so far?

Lookup via hash vs. direct indexing?  Over millions of iterations?  Im
confident the current implementation would win. 

There has obviously been a huge amount of effort invested in getting
axioms domain representation resonably compact and fast.

> | I have not yet seen anything convincing me that a change is essential.
> 
> I was looking more for feedback than convincing you -- I don't think I 
> can convince you of anything.

You might just suprise yourself.  After all, you did manage to
convince me that using a hash as doman representation is a bad idea :)

Your proposing a deep systemic change.  Such things must be torn apart
and analyzed.  Im just attempting to contribute to that process.


Take care,
Steve





reply via email to

[Prev in Thread] Current Thread [Next in Thread]