mife001@MAILSERV.ZDV.UNI-TUEBINGEN.DE (W. Felscher) (05/21/91)
Dear Mr. Hui,
1. yes, I would like to use J on an RS6000
2. I tried to send this to hui@yrlok.ipsa.reuter.com, but
only got as far as itcyyz.ipsa.reuter.com (5.59/smail2.5/11-23-89)
from where there arrived the reply
bad system name: yrlok
uux failed. code 68
550 <yrlok!hui>... Host unknown
3. So I repeat a question which I had intended not ask over
the net. I have asked it previously, with respect to APL,
to some colleagues at IBM without receiving a
satisfactory answer. Maybe in the case of J, which is
just being build, the situation is more perspicuous:
Is - or will there ever be - documentation on the runtime
(in assembler we would count the number of tacts) which
a particular construction or command requires ? Or is the
dynamic behaviour, in particular buffer allocation, so difficult
to predict that no such information can reasonably be
expected ?
The background to this question is, of course, the
teaching of languages such as APL or J to students. It
then seems impossibile to make theoretical comparitions
between the complexities of different algorithms (which
a certain part of theoretical CS is concerned with).
On the contrary, it appears that in APL every knowledge comes
from practical experience with particular implementations.
And this then leads to programming constructions which,
by the average user, can't be justified theoretically
but only by experiencing that they happen to work
well. It IS very satisfactory to have such powerful
instruments as APL and J are - but one would like to know
more about how to 'program' it.
Walter Felscher
WSI fuer Informatik
Universitaet Tuebingen
Auf der Morgenstelle
D7400 Tuebingen, Germanyweg@convx1.ccit.arizona.edu (Eythan Weg) (05/22/91)
In article <9105211421.AA10695@mailserv.zdv.uni-tuebingen..de> mife001@MAILSERV.ZDV.UNI-TUEBINGEN.DE (W. Felscher) writes:
Dear Mr. Hui,
1. yes, I would like to use J on an RS6000
[......deleted..]
I too if it is not too much trouble.
Eythan