Requester problem
10 messages in this thread
I've been trying out a new file requester called "MagicFileRequester V1.0"
by Stefan Stuntz.
One places the MFR icon into the WBStartup drawer. It can replace the REQ
or ASL file requesters. Using CygnusED, it works great. Now in my arexx
program that uses the Arp GETFILE, when I select one file, nothing
happens! When I select more than one file, they get loaded into the
editor. With tracing on, and selecting one file, the Result is 0, or if 4
files are selected, then the Result is 4.
If one calls GETFILE from a wshell window, selecting one file has the
Result show 1. What could be the problem?
On a side note, I was finally able to get my accounting program which is
written in arexx, compiled with the Dineen Edwards compilier. File size
is 589936 bytes.
Can't help with your requester problem, but I am interested in your
compiled program. Have you run any time checks on it? How fast is an
operation compared to the interpreted version? What is your machine
configuration and how fast is your CPU 7Mz – 25 – 30 – 50 ? What settings
did you use to compile? How big is the source?
-Tom
Tom,
In June or July, I have a review of the compiler coming out in
Amazing Amiga. You will find that compiled programs may run upwards of 3
to 5 times faster depending upon the hardware you use and the particular
arexx instructions. Of possible interest is that arrays (compound symbols)
will be substantially faster. If you have a FPU you will find a
substantial increase in math operations. Disk I/O, however, doesn't
increase a great deal.
T.21355
Steven,
I will not see your article, because Amazing isn't sold around here and the
magazine makes ordering it impossible. I liked it, but have been cut off
from it for some time now. Having spoken to you on this board before about
the compilier, I'm sure the review will be highly positive. If you are also
including programming or compiling tips, my lack of being unable to read
the article saddens me. I do have one question that you might be able to
answer> I have talked with Mark over the phone and asked about being able
to distribute the two programs needed to run the compiled code both in the
public domain and in commercal applications. He answered that a yearly
liceanse is available at $10.00. I neglected to ask if that was a flat ten
dollars or is it ten dollars for every program released or sold. Do you
know?
I am interested in the speed because I run at 25Mhz and have seen a 2 to 1
increase in most of my programs. I remember this ratio was higher (the 3 to
5 : 1) on my old 7Mhz bare bones wonder.
-Tom
Tom,
When last Mark and I discussed the "license" I understood that it
would be a one-time fee. I don't recall him mentioning that it would be a
yearly fee, but that may be what they finally decided. I guess we will
have to see the final docs to know for sure. As for the speed increases, I
ran several different types of benchmarks to compare the interpreted code
with the compiled code and speed increases varied from about 1.5 x for disk
I/O to 5 x for some looping and string/array operations. This was on an
accelerated Amiga, so mileage will vary depending upon the hardware
configuration. On average I found a combined increase of about 3 x. If you
have a math co-processor you will find very substantial increases for
floating point math. On my system it was nearly 20 x! Personally, I am
very pleased with the nearly final product. I am mildly disappointed that
the compiler cannot produce code independent of ARexx, but there must be
trade-offs, I'm sure. Because I write ARexx code exclusively for my own
use, I am not overly concerned with the distribution fees. I'm not likely
to be distributing many ARexx macros and I think most people will not be
producing commercial products, so they can as easily distribute the source.
Tom
My A2000 is running on a GVP 33mhz board, with 9 megs of ram. I've done
only one time check last weekend. Starting the program from the prompt, the
interpreted code (209,362 bytes) took 25 sec before the program was ready,
the compilied code (589,936 bytes) took 18 sec.
Spoke to Mark Downing yesterday and the new manual is coming out in a
couple of weeks. It has several expanded sections on how code can be
optimized and better descriptions on all the compilier switches. Currently
I'm just using the defaults that come up in the intuition interface.
T.21381
Ken,
Thank You for your reply. I am interested in speed in acelerated machines.
So far it looks like there is a great deal of increase. I am also over
joyed about the new manual. I have called mark twicew about the compiling
switches. They were a little confusing.
I have down loaded the Magic File requester because of your last message. I
now use it in CED all the time, but am not that up on it yet. Was this the
requester of your origanal question?
-Tom
Tom, I couldn't agree more about the need for better information for making
good use of those compiler switches. Hopefully the new manual will cover
this.
And yes, Magic File Requester was the one that I have problems with in my
accounting program. Still haven't figured it out yet. Sure works well with
CEDpro tho.
You should be able to PowerPack or Turbo Implode it down to 25% of the
compiled size.
Robert, thanks for the tip. I've only heard of PowerPacker, but not Turbo
Implode. Is there much difference tween these two?