On Tue, Mar 20, 2012 at 03:15:41PM +0100, Waldek Hebisch wrote:
> Serge D. Mechveliani wrote:
> >
> > On Mon, Mar 19, 2012 at 06:24:07PM +0100, Waldek Hebisch wrote:
> > > Ralf Hemmecke wrote:
> > > >
> > > > Personally, I tend to agree that there should be a hashtable that does
> > > > not impose SetCategory on its second argument. Of course, one cannot
> > > > output such a table, but maybe the user doesn't want output anyway.
> > > >
> > >
> > > SetCategory is not very important, but BasicType (that is equality)
> > > is crucial.
> >
> >
> > It is desirable to understand how this `=' is used for the second
> > (Entry) argument in the operations `search' and `insert!'.
>
> Not used at all. I meant equality of keys. AFAICS SetCategory
> for Entry is used to define bunch of extra functions which
> are uninteresting to you. In fact, we probably should not
> require SetCategory for Entry and define those functions
> only conditionally.
Yes.
> > For HashTable, I cannot tell what may be (in my example), for example,
> > the cost of `insert!'.
> [..]
> > Meanwhile, I use HashTable. But in future I shall need to rewise its
> > usage, for the thing may ruin the performance.
>
> As I wrote, there are limitations on types of keys. Basically,
> if you have record with more than two components or an
> array inside the keys, then it will not work.
I hope that Key := Integer
is enough for my purpose.
> AFAICS your ConstructedDomain contains Type inside. This
> is tricky case, it will work most of the time but is not
> guaranteed to always work. Basically, as optimization FriCAS
> runtime tries to construct given type only once. This
> is done by keeping a cache of already constructed types.
> But size of cache is limited and if you construct so many types
> that you overflow the cache then FriCAS may construct the
> same type second time. Technically it means that most of
> the time Lisp EQ function will correctly compute equality
> of types, but it type is constructed second time EQ will
> return false and one needs to use more complicated function
> to establish equality.
> [..]
I have HashTable(Key, Entry),
where Key := Integer; Entry := ConstructedDomain.
And from your above notes I conclude that `search' and `insert!'
certainly _will not_ apply equality to any part of ConstructedDomain.
On the other hand, Axiom insists on declaring SetCategory for
ConstructedDomain.
Also I need "=" for ConstructedDomain for the
_needs other than HasTable_.
Its implementation needs to be like this:
build from cd : ConstructedDomain a certain symbolic representation
descr : InputForm, and put cd = cd' == descr(cd) = descr(cd').
This descr is something instead of SExpression.
It may have a considerable cost,
and I hope that `search' and `insert!' for
HasTable(Integer, ConstructedDomain)
_will not_ involve applyig this "=" for ConstructedDomain.
> So it seems that you can not use your ConstructedDomain as
> HashTable keys.
ConstructedDomain is used for Entry, not for Key.
Thanks,
------
Sergei
[email protected]
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/fricas-devel?hl=en.
