127 replies to this topic

[Scratty]

Sorry for spamming your inbox NPS. I'm a bit distracted to day. Oh well.
Perhaps I should communicate here instead.

I was thinking about those txt-files... one solution would be if I send you the rar-file which I've prepared so that all you need to do is to alter those lines of code yourself which you see fit.. then upload the modified rar-file to me or upload to FS yourself. Any preferences? Of course, this is only if you feel you're up for the task, else just say no.

P.S. Never mind the AFX/FX-9860G confusion. I must have mixed them up somehow .

[LordNPS]

Well actually I feel pretty up for the task and I have a few "bugs" that might be errors from my calculator in specific, nonetheless, i would like to check out .
I haven't had much time to play around with FX-GL2 yet, I just uploaded it to my 9860 this afternoon, however as I siad, I did found a couple of bugs.

Right now I do not have the time for a long post(sorry) but tomorrow I'l make and edit and explain my problems:)

About the other thing, no worrys about spaming my inbox , as you might have noticed I too, posesse an gmail acount, and with those the more the better, pratically ilimited space.
You could e mail me the .zip or .rar with everything packed up providing you don't mind, if you haven't done it already (haven't checked out my mail yet).
I'll then change some lines of code that I see fit (color issues)

Sorry for spamming your inbox NPS

And please do not adress me with NPS, you may call me what most of my friends do Sousa. Thing is , I actually like the Lord thing (since I am from an ancient portuguese "lineage" of nobility,even though portugal has been a republic for now nearly 100 years, and it looks cool on on-line gaming) but NPS alone is just plain lame so use Sousa, which is more a little more close than Lord

[Scratty]

Well actually I feel pretty up for the task and I have a few "bugs" that might be errors from my calculator in specific, nonetheless, i would like to check out .

Sure, they're probably just typos since my calc-PC interface doesn't work and I type it in manually.

I haven't had much time to play around with FX-GL2 yet, I just uploaded it to my 9860 this afternoon, however as I siad, I did found a couple of bugs.

Ah, I'm looking forward to hear about them.

Right now I do not have the time for a long post(sorry) but tomorrow I'l make and edit and explain my problems:)

Of course. Take your time.

You could e mail me the .zip or .rar with everything packed up providing you don't mind, if you haven't done it already (haven't checked out my mail yet).
I'll then change some lines of code that I see fit (color issues)

Sure, I'll send it ASAP.

And please do not adress me with NPS, you may call me what most of my friends do Sousa. Thing is , I actually like the Lord thing (since I am from an ancient portuguese "lineage" of nobility,even though portugal has been a republic for now nearly 100 years, and it looks cool on on-line gaming) but NPS alone is just plain lame so use Sousa, which is more a little more close than Lord

Oh, my sincere apologies for my ignorance mi'lord. LordNPS it will be.

[LordNPS]

Oh, my sincere apologies for my ignorance mi'lord. LordNPS it will be.

No need.. just call me Sousa as I said, it's my real surname, now that I mentioned about that "real" Lord thing it is not aprotpriate, it would seam as if I intended people to recognize myself a superior social status or something (wich is not nice, people might get the feeling that I am "full of myslef" of sorts)


Well let's start with this bug and its (hopefull) coreection:
Bug numero uno :
In the program GLINITE , you have in the second line of code :

Seq(0,Y,1,@+not@,1)->list 6

Well this does always return a sintax error, I deleted that Not @ part and it worked just fine, any problem in removing "not @"?

Sorry I had in mind posting a few more bugs , but meanwhile decided to read the e-mail and saw plenty to be done, so sorry for this postpostpone.(and Dr. House is now starting)

And by the way, some of the text files in the zip were all messed up.The Fx-Gl and Readme are not, yet all others are witha simple "end of line" problem, as my notepad, apears not to recognize the square as an enter.

Thinking about it, I'll rather mail about it
I've worked a bit already on the cat thing... I think I'll be ready for a release tomorrow

[Scratty]

Hmm... that's odd, very odd indeed.

You could try replacing the @+Not @ expression with @+(@<>0) if that works better, but it should really work with Not.

Now, the reason for this expression has to do with when @ = 0 and @ = 1. They both produce a single edge element, but @ = 0 also sets the value via X. As you can see, 0+Not 0 = 0+1 = 1, and 1+Not 1 = 1+0 = 1.
Thus you get a one element list in both cases.

Try typing it in by hand if you used your link. If memory serves, when I used my link some years back, I got a syntax error for a entirely correct program. I checked the code multiple times, token by token, and couldn't find anything wrong. Then I replaced the tokens just at the marker where the Syn Error appeared, and that did the trick. This is a very odd kind of bug (there probably is a sound explanation for it) and retyping that line might solve it, unless you've already tried it. Other than that I don't have any further suggestions as of now.

[LordNPS]

Now, the reason for this expression has to do with when @ = 0 and @ = 1. They both produce a single edge element, but @ = 0 also sets the value via X. As you can see, 0+Not 0 = 0+1 = 1, and 1+Not 1 = 1+0 = 1.
Thus you get a one element list in both cases.

Well,Why can't you just type "1" then, if both results are "1" unless you are counting with other values of course 5,6 for example.

About

Try typing it in by hand if you used your link. If memory serves, when I used my link some years back, I got a syntax error for a entirely correct program. I checked the code multiple times, token by token, and couldn't find anything wrong. Then I replaced the tokens just at the marker where the Syn Error appeared, and that did the trick. This is a very odd kind of bug (there probably is a sound explanation for it) and retyping that line might solve it, unless you've already tried it. Other than that I don't have any further suggestions as of now.

I have tryied out tha already, sorry didn't work.
Well, I guess I can always make an If statement ... perhaps it will not be so optimized, nonetheless it might work.

Now other things, just a few problems :
In GLGENGLP " Not Ans=>Goto 0

"not ans" it always returns a Ma Error. (Bloody Calc )

In the program Walk, with the Lists sugested in the documentation,the image output is this :

Well just seams a little bit to messy to be on purpose

oh there was a small typo in 3D Graph too, just replaced f1 with fn1 and it worked fine...
and by the way now I see what you fully mean with surface ploter. this is tottaly difrent from my 3D grapher, it takes a whole lot more of time, but displays the whole surface... mine displays the surface as a grid, so it uses infinitly less points, which alows an easier and faster manipulation, of course the result in not as complete as the one your displays, that sin(x2+y2... gives up a pretty amazing result.

[Scratty]

Well,Why can't you just type "1" then, if both results are "1" unless you are counting with other values of course 5,6 for example.

Yes, it is because you can tell it to produce several edge elements, like 5 or 6 as you suggested. This is how it works. Assume X = 2.5 then:

@ = 0 => {2.5}
@ = 1 => {0}
@ = 2 => {0,0}
@ = 3 => {0,0,0}
etc..

The GLINITV function works in the same manner.

About
I have tryied out tha already, sorry didn't work.
Well, I guess I can always make an If statement ... perhaps it will not be so optimized, nonetheless it might work.

So, did you try exchanging Not @ with (@<>0) in the sequence statement? That didn't help?

Now other things, just a few problems :
In GLGENGLP " Not Ans=>Goto 0

"not ans" it always returns a Ma Error. (Bloody Calc )

I can understand your frustration. This whole thing is making me suspect the Basic code is parsed differently in FX-9860G compared to the CFX-9950G.
Also here... try change it to Ans<>0=>Goto 0, that should work.

In the program Walk, with the Lists sugested in the documentation,the image output is this :
...
Well just seams a little bit to messy to be on purpose

Hm... it appears there is a fundamental difference in how stat draw works between our two machines.
I think it is the skip-line elements in the edge list (those elements with zero in it) that has been rendered somehow.
In the Fill(-statement in GLDRAW, try to change -1 to some even larger number, like -100. I suspect it wont make much difference, but it's worth a shot. You might also want to try to add an extra line under the fill statement: List 4->List 5, that might help. If neither of these hacks works, then it appears that StatDraw actually performs linear interpolation for lines outside the view, just like GLLINE works. This is both bad and good. The good thing is that the boxes doesn't dissapear when they intersect the borders of the screen, the bad thing is that all objects will be connected to each other no matter what you do.

Btw, set AxesOff to get rid of the black lines at the left and bottom of the screen.

oh there was a small typo in 3D Graph too, just replaced f1 with fn1 and it worked fine...

Interresting. In my calculator it is called f1. Seems to be an upgrade in the Basic language.

and by the way now I see what you fully mean with surface ploter. this is tottaly difrent from my 3D grapher, it takes a whole lot more of time, but displays the whole surface... mine displays the surface as a grid, so it uses infinitly less points, which alows an easier and faster manipulation, of course the result in not as complete as the one your displays, that sin(x2+y2... gives up a pretty amazing result.

Oh, thanx
The plotter is on the one hand quite configurable, but on the other hand it isn't very easy to make the clipping planes follow the camera views and x,y-limits, this must be tweaked manually... or alternatively one could make a heuristic for altering the clipping planes with respect to the point of view and the surface parameter limits.
Btw you can increase the step size to produce surfaces with fewer points in them.

[LordNPS]

That <>1 does not work shame.
About the basic, yes there might be some small little changes (there should be plenty of them as casio basic does totaly sucks as we all know, oh well I guess I shouldn't complain, after all, they are some kind of asian people so I guess that the fact that it is in english and similiar to basic is in itself something to rejoyce)

Now About the Gldraw , I use -300, because the draw stat draws until a ceirtan value , I think it is roughly a view window in every direction...
anyways
I am going to work a bit on the .cat then I'll e-mail it to you.
I changed somethings in the documentation, simply removed the bloody coulor issues ( even thought it is not necessary as my calc ignores them, yet this way it will stay cleaner)

[Scratty]

That <>1 does not work shame.

Oops!!! My bad. It should be @ = 0, not @<>0. (Yes, @<>1 won't work either).
That is, in GLGENGLP: Ans=0=>Goto 0
and in GLINITE: Seq(0,Y,1,@+(@=0,1->list 6
Try that, and it'll probably work better .
Seems I forgot to enable my brain back there.

Now About the Gldraw , I use -300, because the draw stat draws until a ceirtan value , I think it is roughly a view window in every direction...
anyways
I am going to work a bit on the .cat then I'll e-mail it to you.

Oh, so you're gonna make cat files instead of g1m-files?
Anyway, it is good that you've found a hack for making GLDRAW work. That's wonderful news!

So, how does the WALK program perform?

I changed somethings in the documentation, simply removed the bloody coulor issues ( even thought it is not necessary as my calc ignores them, yet this way it will stay cleaner)

Sure. I'll run a diff on the files l8r to see what exactly you've changed.

[LordNPS]

I can understand your frustration. This whole thing is making me suspect the Basic code is parsed differently in FX-9860G compared to the CFX-9950G.
Also here... try change it to Ans<>0=>Goto 0, that should work.

That did work

About that problem in GLINITE none of those worked out, but amazingly @+(@=0) does work. (with parenthesis only)

Well I got the CAT files, after spending 10 m copy pasting text, I found out all I had to do was to go to tools, conversion, Fx9860 - AFX and it did it all automaticly.well done myself.
And sorry for the late response ended up in the hospital last night

Edit : And the wlak program performs quite slowly... I don't know what frame rate you get, but I am sure it must be impeditive.I get something like 3spf or perhaps even more... didn't crhonmetered it...

[Scratty]

About that problem in GLINITE none of those worked out, but amazingly @+(@=0) does work. (with parenthesis only)

Yes... it should . The parenthesis must be there or else the Basic interpreter think you mean (@+@)=0. Note that you don't need trailing parentheses. You can save a few extra bytes by instead writing:

Seq(0,Y,1,@+(@=0,1->List 6

Well I got the CAT files, after spending 10 m copy pasting text, I found out all I had to do was to go to tools, conversion, Fx9860 - AFX and it did it all automaticly.well done myself.
And sorry for the late response ended up in the hospital last night.

Hmm.. you know, you don't need to convert to any other format. If the previous format works ok, then we can use that. Only important thing is that it works.

Sorry to hear you've been hospitalized. How serious was it?

And the wlak program performs quite slowly... I don't know what frame rate you get, but I am sure it must be impeditive.I get something like 3spf or perhaps even more... didn't crhonmetered it...

Hehe, I never said it was fast . Well, it does something about 10-12 spf on mine, and you say it is slow with 3 spf *pfft* . Then you should try enabling the GLLINE mode by setting the If 0-statement to If 1. Then we can really talk about slo-mo!

[caspro]

LordNPS, I hope you are well.

--------------------------------------------------------------------

Rasmus, I've downloaded all the fx-gl programs to the calculator.
The 3D surface of sin(X^2+Y^2)/(X^2+Y^2) is very impressive.
The Shapedemo is pretty good too.

I tested in on the 9860 and then on a 9850GC+
The 9850 is much slower.
But the 9860 is great. It looks like you could program Elite to run on it.

I couldn't get everything to work.
The pyramid in shapedemo has lines going to the origin.

In GLGENGLP there are a couple of lines:
r=91=>1245+4545i
Not Ans=>Goto 0
which cause a Ma Error due to Not'ing a complex number

I don't know if LordNPS has already done this but
I've put FX-GLUT into .cat which I can email you.

However I found that FA-124 and 123 and 122 all corrupted the GLTEAPOT
program area. Perhaps they don't like lines that long.

I downloaded fx-gl.cat and fx-gl_ex.cat from your website and found
fx-gl.cat had syn errors a number of times where there was the variable polar 'r'.
It didn't happen at every 'r' which makes me think the cat files were manually
altered after the txt->cat or g1m conversion. You can type r into txt but
not into fxi/cat/g1m after its been converted- the r on the keyboard is not the same
as the variable r even though the fxi/cat/g1m fonts do not show much difference in
appearance. I can email you a modified fx-gl.cat with the r's put back in.
--------------------------------------------------------------------

What follows below is about using Casform to do the conversion.

I think it's best to just use search..replace.. in a text editor
to change unrecognised symbols into CTF.

To get FX-GL to work in Casform:

Search for
&
replace with
\10^

E Exponential operator (wherever appropriate).

This can be replaced using regular expressions.
e.g. 5E5 can only mean the EXP button as E5 with a variable would be a syn error.

Using the editor TextPad (http://www.textpad.com/)
which has regular expressions:

Search for
\([0-9]\)E\([0-9]\)
replace with
\1\\exp\2

Using E for the EXP button is something I could add to Casform
although it would only work for positive exponents:
e.g. 5E-5 could not be distinguished from 5*E - 5.
--------------------------------------------------------------------

To get the program titles in CTF form to make the output easier to read
I used regular expressions in TextPad:

Search for
--.*$
replace with
@@ Program

then

Search for
\(^.*$\)\n@@ Program
replace with
@@ Program "\1"

This changes for example
GLCYLNDR
--------
into
@@ Program "GLCYLNDR"

This doesn't change the output of each program, it just means you can
paste the whole lot into Casform in one go and it is easier to pick out each
program area from the output for copying.
--------------------------------------------------------------------

[Scratty]

Wow! This is quite some feedback . Thanx!

Rasmus, I've downloaded all the fx-gl programs to the calculator.
The 3D surface of sin(X^2+Y^2)/(X^2+Y^2) is very impressive.
The Shapedemo is pretty good too.

Oh, thanx Roy! I'm glad you liked it!
The lookat-parameters got a little complicated so it might be hard for some people to learn from, but at least the app serves as a decent tool for evaulating models, I guess.

I tested in on the 9860 and then on a 9850GC+
The 9850 is much slower.
But the 9860 is great. It looks like you could program Elite to run on it.

The flight simulator, right?
Yeah, at least in theory. It would be a pain to fly a plane, don't you think? .

I couldn't get everything to work.
The pyramid in shapedemo has lines going to the origin.

Hmm.. that is odd. The other models doesn't share this bug do they?

In GLGENGLP there are a couple of lines:
r=91=>1245+4545i
Not Ans=>Goto 0
which cause a Ma Error due to Not'ing a complex number

Hmm.. judging by LordPNS's bugreport about Not Ans above, it seems that 9850 and 9950 can handle logical operations on complex numbers whereas 9860 can't. I think LordPNS has sent me the FX-GL 2.0 release for 9860, but I haven't checked my mail for a while (had my first day at work this monday, so I'm working on adjusting myself to this new situation).
A fix for this is to replace the statement with Ans=0=>Goto 0

I don't know if LordNPS has already done this but
I've put FX-GLUT into .cat which I can email you.

Aha, I think he has, but please send it to me anyway... I'll make a diff between the two and pick the goodies (no offense meant).

However I found that FA-124 and 123 and 122 all corrupted the GLTEAPOT
program area. Perhaps they don't like lines that long.

Oh, how interesting!! . Hmm... annoying too, since that is one of those programs you really want to avoid typing in manually . Is it possible to salvage something from the transfered program, or do you have to type it all in manually?

I downloaded fx-gl.cat and fx-gl_ex.cat from your website and found
fx-gl.cat had syn errors a number of times where there was the variable polar 'r'.
It didn't happen at every 'r' which makes me think the cat files were manually
altered after the txt->cat or g1m conversion. You can type r into txt but
not into fxi/cat/g1m after its been converted- the r on the keyboard is not the same
as the variable r even though the fxi/cat/g1m fonts do not show much difference in
appearance. I can email you a modified fx-gl.cat with the r's put back in.

Oh, I wasn't aware of that! Thanks for pointing that out!
Yes please do that. I've realized that having a job has a certain impact on the timespan for my sparetime .

--------------------------------------------------------------------
What follows below is about using Casform to do the conversion.
---8<--- snip ---8<---

Ahh.. good that you pointed that out. I should put that information in the readme.

[Scratty]

Oh btw, I know I should stop making additional apps for FX-GLUT, but I guess I just can't help myself . Yesterday I made a nifty tube-generating tool. I found that merging cylinders using GLCATOBJ was a waste of resources (lots of redundant vertices and edges), so I wrote this app which takes two lists which contain the segment distance and segment radius respectively and takes variable N (num slices) as arguments. I've called it GLTUBE, and it will probably be available in FX-GLUT 1.1, whenever that will be released. I predict that the function will be quite versatile. It will simplify modeling and you could use it to generate the "raw material" for modeling a face or a hand, or whatever. At least in my theory .

...And yes, the edge generation is optimal.

Here's the code for those of you who have little patience to wait for a future rerelease of FX-GLUT (byte "optimizations" currently disregarded):

File5
Dim List 5->M
'N=SLICES
Rad
For 0->I To M-1
For 0->J To N-1
File5
List 5[I+1]->Z
List 6[I+1]->r
rcos (2piJ/N)->X
rsin (2piJ/N)->Y
If (I+J)=0
Then 0->@:Prog "GLINIT"
1->X:Prog "GLINITE"
Else Prog "GLADDV"
NI+J+1->X:Prog "GLADDE"
If J=N-1
Then NI+1->X:Prog "GLADDE"
IfEnd
IfEnd
Next
Next
For 1->J To N-1
For 0->I To M-1
If Frac (J/2)
Then N(M-1-I)+J+1->X
Else NI+J+1->X
IfEnd
Prog "GLADDE"
Next
Next

Arguments:
segment offset = File5:List 5
segment radius = File5:List 6
tube slices = N


This example produces a rocket nozzle (well almost) which is axis-aligned with the x-axis:

File5
{0,1,1.5,1.7,2.2}->List 5		%offsets
{1,.7,.2,.33,.33}->List 6		 %radiuses
9->N									%slices
Prog "GLTUBE"
-1.1->Z:0->X~Y:Prog "GLTRANSL"	%center about origin
Deg:90->@:0->X~Z:1->Y:Prog "GLROTATE"	%align with x-axis instead of z-axis
Mat RMat TMat X->Mat X

view with the SHAPEDMO app.

Enjoy...

[LordNPS]

Thank you guys for the concern... about that thing on the hopital it was nothing too serous, just got drunk with a couple of friends at my place... well, actually I drank a lot, really much, including a 600? bottle of 35 years old scotch and plenty of other alcoholic beberrages taht were around the house... and smoked cigars which pissed the crap out of my parents, well as I said, We ended all in the hospital in prospects of alcoholic coma...
Nothing to do again at least in the next 2 months...

Sorry for not being around, and not e-mailing you those .cat and .g1m, i will do so right away, they were ready about 4 days ago or something, just haven't had time since

[Scratty]

Well, I don't want to supress the seriousness in the matter but... it is as people say, you're only young once .

---8<---
... and smoked cigars which pissed the crap out of my parents, well as I said, We ended all in the hospital in prospects of alcoholic coma...
Nothing to do again at least in the next 2 months...
---8<---

[Scratty]

Caspro, did you have those cat-files in handy?

[LordNPS]

Did I forgot something in the cat files?
Are there any complains something I might have skiped or forgotten or missdone?
If there are it's cool, you can say so , I have e-mailed the CAT's to you... I assume you have recieved them... right?

[caspro]

Did I forgot something in the cat files?
Are there any complains something I might have skiped or forgotten or missdone?
If there are it's cool, you can say so , I have e-mailed the CAT's to you... I assume you have recieved them... right?

There is a problem with the 'r' variable. See quote:

I downloaded fx-gl.cat and fx-gl_ex.cat from your website and found
fx-gl.cat had syn errors a number of times where there was the variable polar 'r'.
It didn't happen at every 'r' which makes me think the cat files were manually
altered after the txt->cat or g1m conversion. You can type r into txt but
not into fxi/cat/g1m after its been converted- the r on the keyboard is not the same
as the variable r even though the fxi/cat/g1m fonts do not show much difference in
appearance. I can email you a modified fx-gl.cat with the r's put back in.

Rasmus, I sent you files to your hotmail address and they were bounced back,
so I sent them also to your etek.chalmers address.
Perhaps you could confirm which is your working address.

In the meantime I've put a file called rfx-gl.cat on the page:
http://members.lycos.co.uk/rfam/

My calculator collection now includes the Algebra FX 2.0 Plus

Here's a comparative test:


The below results are for the CUBETEST1

Speed:
---------

AFX2+ - Slowest

9850G, 9850GC+, 9970G - all the same

9860G - Fastest


Not Ans (for complex Ans):
------------------------------

AFX2+, 9860 - Ma Error

9850G, 9850GC+, 9970G - works fine


Other Differences:
---------------------

9850G - displays StatGraph3 at top of screen

[Scratty]

Rasmus, I sent you files to your hotmail address and they were bounced back,
so I sent them also to your etek.chalmers address.
Perhaps you could confirm which is your working address.

Ah, ok. I thought you would use my gmail-address. I'll check my etek account. Just for further notice, since I've graduated, I'll try to close my account there eventually. Rather use my gmail account (stated in the header of the FX-GL doc).
I'll check etek .

In the meantime I've put a file called rfx-gl.cat on the page:
http://members.lycos.co.uk/rfam/

Ah nice! Thx. I've already downloaded them.

My calculator collection now includes the Algebra FX 2.0 Plus

Here's a comparative test:
The below results are for the CUBETEST1

Speed:
---------

AFX2+ - Slowest

9850G, 9850GC+, 9970G - all the same

9860G - Fastest
Not Ans (for complex Ans):
------------------------------

AFX2+, 9860 - Ma Error

9850G, 9850GC+, 9970G - works fine
Other Differences:
---------------------

9850G - displays StatGraph3 at top of screen

Oh, so it wasn't just me that didn't succeed in getting rid of that text. Oh, well. If it is only the 9850/9950 that is affected, there isn't much to worry about I guess. It is just a matter of cosmetics.

About that Not Ans thing.. Changing it to Ans=0 might work. If even that doesn't work, either of the following drastic measures might work: (ReP Ans)=0 or Not ReP Ans.

So, A third platform aye? That's neat. Only sad that the AFX2+ is even slower. I only wonder, how much of the API:s did you need to change to make them work with the AFX2+?

Did I forgot something in the cat files?
Are there any complains something I might have skiped or forgotten or missdone?
If there are it's cool, you can say so , I have e-mailed the CAT's to you... I assume you have recieved them... right?

Yes I've got the files alright .
I cannot complain about anything as of now, since I haven't had the time to look at the contents.
I expect to do that tomorrow (sunday). Hopefully, it is ready to be released tomorrow too.
And then we have the CFX-backend. Gotta make a 2.0.1 for that one too. So I think there'll be a 2.0.1 release for both CFX and 9860. Seems to be fair.


It is quite a new situation for me to have a job. So all the sparetime I used to have disappeared all of a sudden . Well I got some sparetime, but not in the same extent as before. So considering me having other hobbies as well, maintaining the code will be a bit crippled for now on.

[caspro]

About that Not Ans thing.. Changing it to Ans=0 might work. If even that doesn't work, either of the following drastic measures might work: (ReP Ans)=0 or Not ReP Ans.

So, A third platform aye? That's neat. Only sad that the AFX2+ is so slow. I only wonder, how much of the API:s did you need to change to make them work with the AFX2+?

The cubetest works on afx without any changes.

But on AFX File5 seems to have to be entered as File 5 when done manually,
but still accepts File5 from a transfer program.

9860 can get File5 from the catalog function, but File 5 from the usual set-up menu.

I've just run
'DISPLAYS THE TEXT "CASIO"
File6
{56,76,43,55,54}->List 2
Prog "GLGENTXT"

changing File6 to File5 as per forum post (File 5 on the afx)

and got the following results:

9850G -immediately says Arg Error and takes me back to line 2 of GLGENTXT
-so I change it to File6 and Run it again
-it takes a second and then says Ma Error and takes me to the Not Ans
-in RUN mode checking Ans, it is 5115+1177i, Not Ans gives 0,
and Not Ans=>Goto 0 does nothing except display Ans which is now zero
-so I change the line to Ans=0=>Goto 0 and Run it again
- it takes a couple of seconds and then says Syn Error and takes me to GLINITV
to Then X->Mat X11 with the cursor at Mat
-in RUN mode checking X it is 0.7+0.7i which is complex so won't go in a Matrix
-I get the idea to clear all lists in all files and then rerun
-it goes Syn Error in GLINITV as above

9850GC - with File5 in the example it behaves like the 9850G
- changeing it to File6 and clearing all files and lists beforehand
it says Dim Error and goes to GLCPYOBJ to line 2 at the -> where it says List 6->List 3

9970G - with File5 it behaves like 9850G,
- changing it to File6 it does the Not Ans Ma Error,
- changing to Ans=0 and rerunning it tuns without error messages but does nothing
except display Matrix Y which only contains a single entry 1
- clearing all files and lists and rerunning, it takes several seconds and then displays Matrix Y

9860G - start by changing to Ans=0, clearing all lists, using File6 in the Example
- all it does is display Matrix Y which is 1
- doing the same but with File 5 in the example gives Dimension Error in GLGENTXT

AFX - start by changing to Ans=0, clearing all lists, using File6 in the Example
- does nothing except say Done

I used the fx-gl.cat and fx-gl_ex.cat from
http://rasmus.anthin...0_programs.html
editing the fx-gl to put \r back in, in a few places
and got fx-glut.txt from
http://81.216.168.22...eng/fx-glut.txt
which I converted to g1m for the 9860 using casform
and then converted g1m to cat using FA-124 and downloaded to other calcs using FA-123.

I guess it will take a while to figure out what's going on.
I'm going to reread this whole thread to see what's been said about making changes
for different models, and also check that the g1m and cat files contain the same data.

P.S. Not related but: the different models use different amounts of memory
for program areas. A new program with no data in it:
9850G doesn't display, but if I remember from the System Memory it is 17 bytes
9850GC+,9970G 17 bytes
afx2+ 24 bytes
9860G 32 bytes

[caspro]

Okay, all those differences between models are illusions, they are solved by:

1) On the GC I had used Ans<>0 instead of Ans=0, and I had a missing -> somewhere

2) By clearing all data before running the program: all lists, all matrices, all variables etc

Now the example:
File6
{56,76,43,55,54}->List 2
Prog "GLGENTXT"

runs the same on all models:
it does nothing except display the contents of Matrix Y which is 1,
except the afx which just says Done.

[Scratty]

Okay, all those differences between models are illusions, they are solved by:

1) On the GC I had used Ans<>0 instead of Ans=0, and I had a missing -> somewhere

Ah, that would be it. Some posts ago, I stated that Not Ans could be replaced with Ans<>0 which is quite wrong .

2) By clearing all data before running the program: all lists, all matrices, all variables etc

Hmm... that's strange. That shouldn't be necessary really. The GLINIT* routines should initialize lists and matrices properly. Perhaps other models except for the CFX ones need to be initialized in a different way.

Now the example:
File6
{56,76,43,55,54}->List 2
Prog "GLGENTXT"

runs the same on all models:
it does nothing except display the contents of Matrix Y which is 1,
except the afx which just says Done.

Yes this is correct. Now the geometry is loaded into memory. Once this is done, you can render the geometry with an application such as SHAPEDMO or use some other routine which sets up the projection matrix and calls GLDRAW or GLFDRAW.
To render, simply run SHAPEDMO and select option 10.

Now, I discovered a typo in FX-GLUT concerning GLTEAPOT. The big matrix chunk should be assigned to Mat X. As it is now, there is a File6 there instead:

....,.8,.6][1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1File6
...

But it should rather be

,.8,.6][1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1->Mat X
File6
...

Now, the teapot should hopefully work for you.

[Scratty]

LordNPS. Only thing I could find that you've missed is in GLLINE, don't recall which line it was but there's no bug or anything, just one byte less in size .

I'll see if I can work more on the CFX files tomorrow and wrap it up for version 2.0.1. Then I'll take care of the 9860 stuff.

[caspro]

Perhaps other models except for the CFX ones need to be initialized in a different way.

I'll investigate it.

Yes this is correct. Now the geometry is loaded into memory. Once this is done, you can render the geometry with an application such as SHAPEDMO or use some other routine which sets up the projection matrix and calls GLDRAW or GLFDRAW.
To render, simply run SHAPEDMO and select option 10.

Thanks. I suppose it would help if I :r-t-f-m:

Now, I discovered a typo in FX-GLUT concerning GLTEAPOT. The big matrix chunk should be assigned to Mat X. As it is now, there is a File6 there instead:

Now, the teapot should hopefully work for you.

Yes, I could see from the .txt file what it should have been, but if you type that into
FA-124/123 or 122 it corrupts it when you save it by deleting that bit and merging the lines.
Do you have any of this software yourself, to confirm that it is not just happening on my computer.

LordNPS: what about you - does the GLTEAPOT get corrupted on your computer ?

Anyway I tried typing it into the calculator manually a couple of times and just got
errors so I gave up and moved on. I'll try again later.
It may have been the trouble with initialising all lists,mats,vars again.

[LordNPS]

Well I don't know if I fully understand your problem about the GLTeapot, but it works fine on my calculator...
No file corruption nor anything ....

LordNPS. Only thing I could find that you've missed is in GLLINE, don't recall which line it was but there's no bug or anything, just one byte less in size

Well that's weird... it should contain a some bytes less as I removed 2 color lines related lines...
1 Byte eh? when I found myslef beholdding such an astonishing number of bytes I got confused...1 byte, does not really make much sense... oh well...

[Scratty]

I'll investigate it.

Thanks. I suppose it would help if I :r-t-f-m:

Haha, don't worry. I do that myself all the time .

Yes, I could see from the .txt file what it should have been, but if you type that into
FA-124/123 or 122 it corrupts it when you save it by deleting that bit and merging the lines.
Do you have any of this software yourself, to confirm that it is not just happening on my computer.

I use one of the oldest versions, FA-122 or something. Since my moms computer runs win98 and the l8r models didn't seem to work on that OS. (My computer uses Ubuntu and wine + FA-122 doesn't appear to work for me).

Now, the reason I've missed all those r symbols is because I've initially manually typed in everything into FA-122. Then when I discovered that your casform work as a charm after rebooting the computer, I used that instead for the FX-GLUT and FX-GL_EX files, hence the lack of r-errors in those. I will use casform henceforth (it speeded up the workflow at least tenfold).

Anyway I tried typing it into the calculator manually a couple of times and just got
errors so I gave up and moved on. I'll try again later.
It may have been the trouble with initialising all lists,mats,vars again.

Hmm... I'm sad to hear that .
Just thinking, perhaps I should go back to the version of the teapot that uses GLADDV/GLADDE calls instead. It is much larger, but at least it will hopefully transfer well to the calc.

Well I don't know if I fully understand your problem about the GLTeapot, but it works fine on my calculator...
No file corruption nor anything ....

Hmm.. perhaps you used another version of FA-12* to upload it?

Well that's weird... it should contain a some bytes less as I removed 2 color lines related lines...
1 Byte eh? when I found myslef beholdding such an astonishing number of bytes I got confused...1 byte, does not really make much sense... oh well...

In GLLINE:

\If D<0
\Then E-D/(A-D)*(B-E\->E
0\->D
\IfEnd

It should look like this:

\If D<0
\Then E-D*(B-E)/(A-D\->E
0\->D
\IfEnd

This is of course as you say, of no concern, it just looks more.. hmm... consistent .

[liquid]

Hey id just like to say i really like what your doing here.
I honestly never thought anyone would get real 3d going on the old casio basic calcs.
I tried once, but got a little too confused with all the matrix algeba, but i had never studied matrix algebra back then, i was a newbie at all of it.
Its sorta sad in a way thought, that we have too program in such slow and limiting casio basic, but in some ways thats what makes it a real challenge and fun when you get something working.
It would've been so cool if they allowed assymbler for the cheap calcs, I was really dissapointed when i first tried to program in casio basic, i remember seeing it draw at one pixel per second, and it broke my heart!

[Scratty]

Hey id just like to say i really like what your doing here.

Thanx! It's very nice to hear some positive feedback. Good for my programming mojo .

I honestly never thought anyone would get real 3d going on the old casio basic calcs.
I tried once, but got a little too confused with all the matrix algeba, but i had never studied matrix algebra back then, i was a newbie at all of it.

Ah, well... I was playing with the idea and concluded that it would be very much doable. I made an attempt to make another doom clone in true 3d one year ago, but didn't get it right. I guess it was because 3d programming was new for me then. After having taken the sequel course in computer graphics at my university, my knowledge in the subject had matured somewhat and I felt I should make another attempt, and this is the result. Of course, to my aid I had the reference pages for GLU and GLUT and the OpenGL website.

Actually, before I decided doing FX-GL, I had another project going which was a raytracer for Casio calcs, which I have written about earlier.

My colleague at work suggested that FX-GL should use some simple shading, like normal flat diffuse shading or so, but we both immediately realized that it would be too slow for a semi-realtime application. In such cases shading is required - like diffuse and specular - the raytracer I did is more appropriate. It can be extended to work on other primitives and to use refraction, etc. Only battery life time and ones patience sets the limit in that case .

Its sorta sad in a way thought, that we have too program in such slow and limiting casio basic, but in some ways thats what makes it a real challenge and fun when you get something working.

Absolutetly. On the one hand, programming basic sucks in performance, but on the other hand making a program super-optimized in basic is quite rewarding. However, the reward doesn't quite pay off for the price of lack in performance, naturally.

It would've been so cool if they allowed assymbler for the cheap calcs, I was really dissapointed when i first tried to program in casio basic, i remember seeing it draw at one pixel per second, and it broke my heart!

I absolutely concur! It broke my heart too.
We should start a manifesto for incorporating assembly support in newer calcs.
(reservation: I'm not aware of the status of current calculator development.)

Once again, thank you for your comments Liquid!


This weekend, I'll be away launching my rocket with some other guys, just to keep the hobby alive. Also, the graphics card from my ol' computer has cored, so I cannot do any work/update on FX-GL & c:o in the near days. I apologise for the inconvenience.

[LordNPS]

This weekend, I'll be away launching my rocket with some other guys, just to keep the hobby alive. Also, the graphics card from my ol' computer has cored, so I cannot do any work/update on FX-GL & c:o in the near days. I apologise for the inconvenience.

Thats great what class of rocket engine? Solid fuel?
Here in portugal amateur rocketry is rather opressed by stupid laws... we cant use rocket engines above the G class...
Which sucks, my father and I have been building a pseudo "scale" rocket of the Soyuz TM.31 at a small hangar in an aviation field nearby, it is desproportionate though, as you may well know, in rocketry, things aren't proportional, it has a much smaller engine thrust area, and does not use the standard 5 * 4 small engines of the soyuz... we thought doing that however it would be much expensive, because you would need 20 class E engines wich are only a little cheaper than class G.
It uses 5 G80's from aerotech.
For it to work properly we would need 5 K engines which as I state are not allowed here, nontheless we are using a little trick, in the fuel output which allows us the G80's long period duration, and allows a pretty start Impulsion, which is to use 2 aditional E4 from aeorotech as well, wich provide plenty of purpolsion for the period of time that the G engines do not, then they are discarded, as they burn out...
It is a big boy, that rocket
I bet it's gonna blow on launch

The main problem is the sincronization of the engines, for the rocket to go straight up...

Of course that this project is not only mine and my father's, we have 2 other rocketrist helping us.

[Scratty]

Thats great what class of rocket engine? Solid fuel?

We just launched LPR rockets from A to D that is. I had an E motor too. We were three guys, and we launched 28 rockets altogether. One guy was a total newcomer but nontheless had no bad luck at all (in contrast to the other guy who lost two rockets). In our society SAMRA (which isn't quite so active anymore) some people launch HPR rockets.. that's H and upwards. The highest impulse we've used was an M class motor. Some of us use hybrids.

Here in portugal amateur rocketry is rather opressed by stupid laws... we cant use rocket engines above the G class...

Yeah I can understand that. Then we also have the new EU-regulations. Engines have to be CE-certified and that process seems to be very slow.

Which sucks, my father and I have been building a pseudo "scale" rocket of the Soyuz TM.31 at a small hangar in an aviation field nearby, it is desproportionate though, as you may well know, in rocketry, things aren't proportional, it has a much smaller engine thrust area, and does not use the standard 5 * 4 small engines of the soyuz... we thought doing that however it would be much expensive, because you would need 20 class E engines wich are only a little cheaper than class G.
It uses 5 G80's from aerotech.

Mmm.. clustering is nice. Never tried it though. We talked a little about it yesterday. Think I'm gonna try on multistaging next though.
5 G80's? Wow! That's nearly an I-motor!

For it to work properly we would need 5 K engines which as I state are not allowed here, nontheless we are using a little trick, in the fuel output which allows us the G80's long period duration, and allows a pretty start Impulsion, which is to use 2 aditional E4 from aeorotech as well, wich provide plenty of purpolsion for the period of time that the G engines do not, then they are discarded, as they burn out...
It is a big boy, that rocket
I bet it's gonna blow on launch

From the sounds of it,... yeah! . Impressive!

The main problem is the sincronization of the engines, for the rocket to go straight up...

Of course that this project is not only mine and my father's, we have 2 other rocketrist helping us.

Yep synching ignition and pressure buildup is probably the hardest bit when clustering. Hope you'll get it up and flying . You must show a video on the launch l8r!

---

Hmm... I'm been thinking about FX-GL. I think I will have to split the package up in three parts. It just won't work having FX-GL, FX-GLUT and FX-GL_EX in the same distribution.

I've only got 6 hours sleep each night, hence I'm quite tired after this week's work. I'm not really used to this. Oh, well. I'll try to be more productive today, but can't make any promises though.

[caspro]

Now that you've rocketed back to the casio:

About converting txt -> cat:

The GLGENGLP routine contains quite a few complex real part: ReP's.

Casform usually took lower case e to mean exponential e^ if it is not adjacent to other lower
case letters. There are a couple of exceptions but I missed the exception ReP.

So new version 9, takes care of this, as well as E where appropriate e.g. 5E5.

[Scratty]

Yes, I noticed this. I altered the original cat file guided by your changes, so there wasn't any problems with that for my part. Thanx for the heads up anyways!

Nice feature with the exponential! I'll download the new revision asap.

[Scratty]

There you go. The latest version of FX-GL, FX-GLUT and examples for both CFX and FX-9860G platforms are now released!
Sorry for taking so long (rearranging files and documentation took some time).
There's not much updated in these revisions, hence the patch-numbers.

I'll soon add GLTUBE, GLANIMMK and GLANIMPL to the FX-GLUT API. The two last programs are used for making animations using pictures (quite simple really).

Just curious to know, how many frames or pictures are available in the FX-9860G calculator?

Please let me know if you notice any peculiarities, oddities or bugs in these relases.

[LordNPS]

Nice to hear that

Lately I haven't been paying much atention to the topci (well, we are a few steps closer to a SDK, we now praticly everything we need).

9860, sadly is capable of only 20 simultaneus pictures, however you can store them in the flash memory, and then transfer them to the picture memory in RAM, which is kind of nice...
I don't know any way to interact with the file system through Basic though

[Scratty]

20 pictures huh? Thats more than I have. I only have 6 pictures . Guess you could animate some neat stuff if you keep the frame rate to a moderate level.

Btw, I've finished working on the GLARCBAL program (GL ArcBall). Now you can interactively rotate the scene by "clicking" at two points on the screen. Only limitation is that you have to define your up-vector as (0,1,0) and eye position along the z-axis. You might be able to bypass this somehow though.

Now, I've thought about making a GLUNPROJ program too (GL UnProject) which enables you to click on points in view space and locate the corresponding vertices. This opens up the possibility to make a 3D modeling application. Now, the problem is, how on earth do I get the screen z-coordinate?

glReadPixels(winX, winY, 1, 1, GL_DEPTH_COMPONENT, GL_FLOAT, &winZ);

As far as FX-GL goes, yes I think it is pretty much complete (Ok, a glViewport would be useful). Regarding FX-GLUT though, there might still be a few programs that could be added to it.

[Scratty]

Wohoo! It was way easier than I expected to make a GLUNPROJ. It is just a matter of iterating through the (x,y) values in the Y-matrix (the projected model coordinates) and checking how near each point you clicked (you can programatically enter screen coordinates too).

GLUNPROJ:

Dim Mat Y
List Ans[2
For 1->Z To Ans+1
Z<=Ans=>(X-Mat Y[1,Z])^2+(Y-Mat Y[2,Z])^2<=r^2=>Break
Next
Z=Ans+1=>0->Z
Z->@

where @ is the index of the vertex that match the screen coordinate. If @ = 0, then no matching vertex was found.
r is the tolerance radius.

In your application:

Plot _
2->r:Prog "GLUNPROJ"

[Scratty]

Sorry for spamming the thread.
Just wanna say that I've made a modeling application now. I've chosen to call it 3DFX . You can zoom in/out the scene, you can pan the scene and of course rotate the scene (uses the ArcBall program). The world coordinate axes is displayed in the upper right corner of the screen (blue=x, green=y, orange=z). You can switch between default camera views such as front, back, left, right, top, bottom (this resets the arcball rotation state).

The only actual modeling feature available is the possibility to edit single vertices. Just click on a vertex (within a two pixels range) and the vertex number along with its coordinates will be displayed in text mode. Then you'll be asked to enter new vertex coordinates. In order to preserve coordinate values simply type X, Y or Z whenever appropriate.

Since vieports are not supported by FX-GL just yet, you'll have to live with one main view of the scene .
Therefore, it'll be a perspective projection for that view (it can rather easily be modified to use ortho view instead. Just mind the zooming!).

Next feature will be the ability to scale vertices around some centre point or so.

Ray tracing isn't supported yet .

If you use the GLTUBE program posted earlier in this thread, you have an IMHO a quite powerful tool to model just about anything you can think of. This is the intended workflow:

• Load the raw object (the tube) into memory using GLTUBE, or load any other primitive or primitive compound.
• Modify the tube-object using 3DFX.
• If you want to save the geometry data, either:
• Write down the contents from the X-matrix on a piece of paper (and optionally put the data into a new file in your calculator).
• Transfer the data to your PC.

Btw, if I decide to post this app here on the FS, it'll be posted in its own file.

[Scratty]

Hey! What ever happened to the FX-9860G counterparts of FX-GLUT and FX-GL_EX?
They seem to be gone from the FS.

For those of you who want FX-GLUT and FX-GL_EX for the FX-9860G can take a look at:
http://rasmus.anthin...0_programs.html
Check at the near bottom of the page.

[Scratty]

Ok. If no one gives me a reason why GLUT and GL_EX has been removed, I will assume they disappeared due to some glitch in the FS scripts and thus I'll attempt to upload again and hope it'll work this time.

If any of you admins deleted them due to having the same name as those archives available in the CFX folder, then you were presumtious. The archives do differ even though they are very similar. For example, logic operations and colours are handled differently.