> < ^ Date: Mon, 01 Mar 1993 20:35:40 +0100
> < ^ From: Martin Schoenert <martin.schoenert@math.rwth-aachen.de >
< ^ Subject: Re: First impressions of 3.2
In his e-mail message of 1993/02/23, Jean Michel writes
    I have just installed 3.2 on my PC (a 66Mhz 486). The number of
    gapstones has jumped from 16000 in 3.1 to 28000 in 3.2! Since the
    improvement seen on the Sun sparc SLC is just 13000 to 16000, I conclude
    that much of the improvement on the PC must be due to a better
    C compiler. Impressive! (and congratulations!)

As Frank already noted, the main reason for the performance jump is that
GAP now checks for user interrupt less frequently.

He continues:

Also I like very much the facility offered by {} indexing, which
I timed as being 5 times faster than Sublist. However, I find
myself writing all the times things like:
l{[4..7]} or l{[3,5,9]}
could not a syntax like
l{4..7} or l{3,5,9}
be accepted to mean the same?

We are giving some thoughts to extend the concept of lists to the more
general concept of a table. In a table the index could be any object,
not just a positive integer. So you could, for example, index into a
table with permutations, lists, etc. At one time the syntax I had in
mind for sublist extraction made it neccessary to always have a list
expression in the sublist construct '<list>{<poss>}', so that GAP could
distinguish between extracting a single element at an index given by a
list and extracting a whole sublist at indices given by a list. The
current syntax cannot lead to such a confusion, so 'l{4..7}' or
'l{3,5,9}' could -- and probably will -- be accepted in a future version.

He continues:

One last thing: I notices that 3+[] and []+3 are now both accepted and
return [].

Yes. But you probably should not use this. Or at least if you use this
be aware that further rewrites of the list packages may remove this
funcionality again (though this is not very likely). I will write more
on the subject of empty lists in a seperate e-mail message.


-- .- .-. - .. -.  .-.. --- ...- . ...  .- -. -. .. -.- .-
Martin Sch"onert,   Martin.Schoenert@Math.RWTH-Aachen.DE,  +49 241 804551
Lehrstuhl D f"ur Mathematik, Templergraben 64, RWTH, D 51 Aachen, Germany

> < [top]