Testers required for the mouse modules of the TC-Lib

[Jark]

I managed to build a set of mouse modules in SVGA true-colour modes. It works OK on my PC, but Antoni has reported to me that these modules don't work on his machine: I suppose it will or won't work from one machine to another...

Can you folks have a look at :

http://mandelbrot.dazibao.free.fr/Tclib/Tclib15.zip

and help finding what I need to add to get module that always work ?

Here's a screenshot of what this lib can do (it's an old pic, but it is close to what the current version looks like)

[ak00ma]

The mouse routines worked for me...but i had to de-call the Ffix sub to get it work:

Code:

'========================================================================

CONST Pi = 3.14159265358979#         ' For angles management
CONST SaveDir$ = "C:\Tclib\"         ' For files management
CONST eps = 1.401298E-45             ' Avoids logarithm overflow
REM Ffix                                 ' Floating point patch
' ==============================================================
MainControl
EndProg

and after a while there occurred a "Division by zero" error...probably because of the de-calling of Ffix

[Antoni Gual]

Jark:
I think all my problems with the mouse are W2000's fault. I found the y coord of the mouse is limited to 199 by some strange reason. I re-checked my old 3-d mesh rotator in SVGA where I was adding mouse support and it has now the same problem (It worked ok under W98).
In a mouse demo in screen 12 I can read y coords of 480 without any problem...

[Jark]

I had a lot of trouble with the mouse coords in the beginning, and it finally worked without really knowing why... The mickey stuff is still mysterious for me: I made that work, but don't ask me how !

The most surprising is that the mouse management should not be dependant on QB : the interrupt functions are purely hardware and BIOS related, and the functions are the same in C, Pascal or whatever. It is slow with QB, and that's all...

The best I can do is to extract from the lib the minimum set of subs to allow testing that on various configurations...

[Jark]

I did not take time to check the lib in detail with all the versions of QB.

It works fine with QB4.5, and seems OK with QB7.0

As for QB4.0, I have some double definition error messages: since I gave up using QB4.0 after discovering ffix, I do not want to spend time on it.

I know I had some trouble with the heavy raytracing modules (Shape3 and Shape4) in some cases (mainly, impossible to compile the lib into an exe file). I even had some 'division by zero' errors on lines which had no division!

What you can do is to remove Shape3 and Shape4 to check the lib.

Recommended configuration is : win95 or win98, and QB45 with a merged library (qb.qlb+ffix). Of course, a fast machine is more confortable...

I really need to split the lib into modules to get rid of all these errors: maintenance will be much easier!

[oracle]

Wow, Quintuple post! That's gotta be a record! :rotfl:

[Jark]

I will try to clear that... I post through the intranet of my customer, and it's a mess!

[oracle]

Yo wildcard, you *do* still exist eh? You can fix that, can't you :wink: .

Or do you not bother to moderate when you see someone responsible starting the posting. Good for you!

[Dav]

There. It's Fixed now.

Jark, great LIB you have going there. Really cool. 8)

- Dav

[oracle]

TC-Lib is goin' on my site... but maybe not at the launch, I want to do a proper review etc...