zrzm0111@helpdesk.rus.uni-stuttgart.de (MUFTI) (06/02/91)
Hi !
Here is a "new" argument in languagewars:
Most common opinion after the langague-war was, that you should normally use
highlevel languages and should use assembler for critical parts of the program,
cause highlevel constructs like "x=x+2*(y-34)+z" are faster to write and
very much easier to maintain than the same in Assembler.
Most agree, that the misadvantage was, that the librarys lacks some memory for
unused routines.
There is a way to use mathematical strings like "x=x+2*(y-34)+z" in programs,
which run only through a macro-assembler producing a absolutefile and don't
lack so much memory for the libraries.
I wrote a little program, that converts such mathematical strings in assembler-
macros, but it's not really ready yet.
It supports only Integer-calcuation ( cause F. Lancester's update for TLA
supporting Floating-point-coprozessor-command isn't ready yet ) and the
quick hacked division-macro is too slow ( this is the next thing I do ).
Before I spend much brains in a better optimisation of the target-code,
( currently temporaery variables are hold in memory 8-{ ), I want to know
if the are some errors.
Can someone test it ?
I build a little demo, to declare it's use.
( Be warned: Cause gawk is the only FREEWARE-tool for archimedes under RISCOS,
which supports high-level-language constructs and is full portable to most
other machines, I wrote the program mainly with it. Cause gawk is made for
faster machines, the program runs very slow. Cause it's developped under UNIX,
it must be started from the Command-line-Interpreter and therefore things like
desktop-interfaces and multitasking is impossible (getline <"kbd:"; doesn't
work correct (bug ?) so it's quite complicated to read a string from
keyboard .... ) )
The program is called !formula and it's declared as FREEWARE.
The demo (and the program !formula) is available via ftp from rusmv1.rus.uni-
stuttgart.de ( and therefore via mail by bitftp@pucc.princeton.edu ).
File is /soft/acorn/riscos/programming/macro-asm/formula-demo.
You need TLA (comp.binaries.acorn or rusmv1) and
gawk (info-server@newcastle.ac.uk or rusmv1) to use it.
My opinion about new ARM-assembler-commands:
There was some rumors about new ARM-assembler-commands.
The example of the extentions of the 8086-chip to the 80186-chip, which
made new assemblercommands in real-mode, which nearly not used by any
MSDOS- software, show, that you have to think to downward-compatibiltiy of the
new commands.
The only way I see, to do this for the archimedes, is to put a software
emulator for this new commands ( which behaves like the FPU-emulator ) into
the public domain.
Questions:
1. Is there a tool like !lineedit, which can be used without starting the
desktop at all ?
2. Can someone explain me what filetype the file !arc100.arc100 has ? I get
funny results with fileinfo !arc100.arc100.
3. Someone posted, he is angry by extracting .tar.Z -files cause he have to
give lots of CLI-commands. Why don't we declare a standard (like
!submit/!extract for .tar.Z.uue -files) ?
4. Same procedure as every time:
Does everyone knows implementations of PD/Shareware/FREEWARE Editors ?
We want to public them on our ftp-server in Stuttgart.
so long
MUFTI
ps:
Cause G.Toal find out, that mail through the connections between janet
and internet may disappear, I will from now on send back a short reply
to all mails.
If the problems hold on, I have to change the mainly used host ...
internetadress:
zrzm0111@helpdesk.rus.uni-stuttgart.de
which is for janet (G.Toal told me that):
zrzm0111%de.uni-stuttgart.rus.helpdesk@earn-relay
zrzm0111%de.uni-stuttgart.rus.helpdesk@nsfnet-relay
bitnetadress:
ZRZM AT DS0RUS1I