quesa-develop — Mailing list for Quesa development

Re: Zero length vectors causing problems

[Tom S. <to...@as...>]

> example:
> /*
> unsigned original_mask = _controlfp(0,0);
>    unsigned mask = original_mask;
> 
>    mask|=(_EM_ZERODIVIDE);
>    _controlfp(mask, _MCW_EM);
>    */
> CPrintDialog pd(FALSE);
>          pd.GetDefaults();
> 
>          float a,r;
>          a=0;
> 
>          r=1/a;
> 
> If I set HP laserjet 1100  like default  program crash. If  I canc /* and 
> */ program run fine.
> With other printer I have not problem.See the documentation of _controlfp. 
> Try my little sample... is more simple solve this.
> All only in WIN 98!!


Yes it was the printer driver for me too.

Your solution sorted it out.

Thankyou very much Daniele!

Tom

Re: Should there be one renderer for each view?

[Dair G. <da...@re...>]

James W. Walker wrote:

>I wasn't asking about multiple renderers on one view, but rather=20
>multiple views on one renderer.

Ah, OK. I don't think that will have been a restriction before, although
I suspect things wouldn't have worked if two views were submitted to at
the same time.

I.e., if two views both share a renderer then this will work:

    start rendering  to view A
    submight objects to view A
    finish rendering to view A

    start rendering  to view B
    submight objects to view B
    finish rendering to view B

But this won't:

    start rendering  to view A
    start rendering  to view B
    submight objects to view A
    submight objects to view B
    finish rendering to view A
    finish rendering to view B

Renderers are probably going to want to store some per-rendering-loop
state in their instance data (e.g., batching up transparent triangles)
that would break in this case.

You could make it work of course - renderers would just need to separate
their "this is my state" vs "this is my rendering state", and have the
latter stored in an array indexed by the current view or something...


>I don't actually want to do that. I've been thinking about possible
>solutions to bug 902975, "Use shared texture namespace".  If I could
>assume that there is a one to one correspondence between views and
>renderer, it would make the problem a little simpler.

Could we just keep a proxy GL context around at all time? That seems to
be the approach they're talking about in QA1031 - something like:

  - texture manager creates a fake gl context on demand
  - texture manager creates/updates gl textures for qd3d textures
  - renderer creates contexts that are shared with the TM context
  - texture manager disposes of fake gl context on shutdown

Problem is the texture manager object needs to existing independently of
the renderers - not sure what the best way to handle that is, perhaps
when the renderer library is loaded/unloaded so that it's always around
even as individual renderers come and go.


-dair
___________________________________________________
mailto:dair+refnum.com       http://www.refnum.com/

Re: Should there be one renderer for each view?

[James W. W. <os...@jw...>]

On Nov 14, 2004, at 6:53 AM, Dair Grant wrote:

> James W. Walker wrote:
>
>> I don't see anything in the QD3D manual saying that you can't attach
>> one renderer to multiple views, although I don't think I've ever seen
>> sample code that did that.  Also, the Quesa implementation of the
>> interactive renderer seems to assume that it is being used for one 
>> view
>> (and hence one draw context), since its instance data contains some
>> OpenGL state.  Is it safe to assume that a renderer will be used by at
>> most one view?
>
> Yep - the QD3D model is really that rendering happens "to a view", 
> where
> that view has a single camera (where to draw from), draw context (where
> to draw to), and renderer (what to draw with).
>
> Is there a situation you're thinking of where you would like to have
> multiple renderers attached at once? (executing in series I could see,
> in parallel I'm not sure how it would work)

I wasn't asking about multiple renderers on one view, but rather 
multiple views on one renderer.  I don't actually want to do that.  
I've been thinking about possible solutions to bug 902975, "Use shared 
texture namespace".  If I could assume that there is a one to one 
correspondence between views and renderer, it would make the problem a 
little simpler.
--
<http://www.jwwalker.com/>

Re: quesa.org

[Dair G. <da...@re...>]

James W. Walker wrote:

>On Nov 12, 2004, at 12:37 AM, Dair Grant wrote:
>
>> OK, that's set up - I've added entries for quesa.org and
>> www.quesa.org. It says they take up to 6 hours for the changes to
>> take effect, so I'll check again later today and see if their status
>> has changed (right now they just say "scheduled for creation").
>
>Cool.  But in case I was unclear, I don't think that the DNS
>information is set up as instructed yet.

It looks like <http://quesa.org/> does now respond correctly, although
<http://www.quesa.org/> gives the empty apache page.

There's no extra status info you can get to for the VHOST entries on
sourceforge.net itself, so I think all we need is for the "Have CNAME
Records Set-Up" section at:

  <https://sourceforge.net/docman/display_doc.php?docid=3D777&group_id=3D1>

Mike, am copying you as you're not on the list - does that page make
sense from your point of view?


-dair
___________________________________________________
mailto:dair+refnum.com       http://www.refnum.com/

Re: Should there be one renderer for each view?

[Dair G. <da...@re...>]

James W. Walker wrote:

>I don't see anything in the QD3D manual saying that you can't attach=20
>one renderer to multiple views, although I don't think I've ever seen=20
>sample code that did that.  Also, the Quesa implementation of the=20
>interactive renderer seems to assume that it is being used for one view=20
>(and hence one draw context), since its instance data contains some=20
>OpenGL state.  Is it safe to assume that a renderer will be used by at=20
>most one view?

Yep - the QD3D model is really that rendering happens "to a view", where
that view has a single camera (where to draw from), draw context (where
to draw to), and renderer (what to draw with).

Is there a situation you're thinking of where you would like to have
multiple renderers attached at once? (executing in series I could see,
in parallel I'm not sure how it would work)



-dair
___________________________________________________
mailto:dair+refnum.com       http://www.refnum.com/

Should there be one renderer for each view?

[James W. W. <os...@jw...>]

I don't see anything in the QD3D manual saying that you can't attach 
one renderer to multiple views, although I don't think I've ever seen 
sample code that did that.  Also, the Quesa implementation of the 
interactive renderer seems to assume that it is being used for one view 
(and hence one draw context), since its instance data contains some 
OpenGL state.  Is it safe to assume that a renderer will be used by at 
most one view?
--
<http://www.jwwalker.com/>

Re: quesa.org

[James W. W. <os...@jw...>]

On Nov 12, 2004, at 12:37 AM, Dair Grant wrote:

> OK, that's set up - I've added entries for quesa.org and www.quesa.org.
> It says they take up to 6 hours for the changes to take effect, so I'll
> check again later today and see if their status has changed (right now
> they just say "scheduled for creation").

Cool.  But in case I was unclear, I don't think that the DNS 
information is set up as instructed yet.
--
<http://www.jwwalker.com/>

Re: quesa.org

[Dair G. <da...@re...>]

James W. Walker wrote:

>After the DNS information is set up appropriately, I think Dair needs=20
>to configure the VHOST.  At least, I don't have the privileges to do=20
>it.

OK, that's set up - I've added entries for quesa.org and www.quesa.org.
It says they take up to 6 hours for the changes to take effect, so I'll
check again later today and see if their status has changed (right now
they just say "scheduled for creation").


-dair
___________________________________________________
mailto:dair+refnum.com       http://www.refnum.com/

Re: quesa.org

[James W. W. <os...@jw...>]

On Nov 11, 2004, at 6:53 AM, Joseph J. Strout wrote:

> I think it'd be better to pursue the DNS/SourceForge approach a bit 
> further first, though this is a reasonable fall-back plan if all else 
> fails.

Check out this page: 
<http://sourceforge.net/docman/display_doc.php?docid=777&group_id=1>

After the DNS information is set up appropriately, I think Dair needs 
to configure the VHOST.  At least, I don't have the privileges to do 
it.
--
<http://www.jwwalker.com/>

Re: Zero length vectors causing problems

[Daniele C. <dca...@in...>]

example:
/*
unsigned original_mask = _controlfp(0,0);
   unsigned mask = original_mask;

   mask|=(_EM_ZERODIVIDE);
   _controlfp(mask, _MCW_EM);
   */
CPrintDialog pd(FALSE);
         pd.GetDefaults();

         float a,r;
         a=0;

         r=1/a;

If I set HP laserjet 1100  like default  program crash. If  I canc /* and 
*/ program run fine.
With other printer I have not problem.See the documentation of _controlfp. 
Try my little sample... is more simple solve this.
All only in WIN 98!!


At 15.15 11/11/04, you wrote:

>----- Original Message -----
> >From: "Daniele Cavallini" <dca...@in...>
> > I had some problem.
> > My problem was caused by driver of printer. There was a series of HP 
> drivers that were a bit buggy and by that I mean they turned
>off the floating point exception handling and never returned the state.
>
>Thanks, Daniele.
>
> > You might want to save the state before the call, then return the state 
> after the call.
>
>Save the state of what? Could you explain how I would do this?





> > You can try if is this problem add a new printer and set this new 
> printer like defaut (for example print pdf).



>Ok, will do. I might also try deleting all printer drivers and see if that 
>helps
>
>Tom
>.
>
>
>
>-------------------------------------------------------
>This SF.Net email is sponsored by:
>Sybase ASE Linux Express Edition - download now for FREE
>LinuxWorld Reader's Choice Award Winner for best database on Linux.
>http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
>_______________________________________________
>Quesa-develop mailing list
>Que...@li...
>https://lists.sourceforge.net/lists/listinfo/quesa-develop

Re: quesa.org

[Joseph J. S. <jo...@st...>]

At 4:44 PM -0800 11/10/04, Mike Wheeler wrote:

>The following URLs should all be equivalent if not for some unusual 
>server configuration on the SourceForge side. Unfortunately only one 
>of them works.
>
>http://www.quesa.org - 404 Not Found
>http://66.35.250.209 - 404 Not Found
>http://quesa.sourceforge.net - Quesa homepage

That is really strange.  I think you should try asking SourceForge 
tech support about it -- I've found them to be very responsive and 
helpful in the past.

>It looks to me like to make this work quesa.org will have to come 
>back to being an Artifice hosted server which issues a redirect to 
>http://quesa.sourceforge.net.

I think it'd be better to pursue the DNS/SourceForge approach a bit 
further first, though this is a reasonable fall-back plan if all else 
fails.

Thanks,
- Joe

-- 
,------------------------------------------------------------------.
|    Joseph J. Strout         Check out the Mac Web Directory:     |
|    jo...@st...           http://www.macwebdir.com/            |
`------------------------------------------------------------------'

Re: Zero length vectors causing problems

[Tom S. <to...@as...>]

----- Original Message ----- 
>From: "Daniele Cavallini" <dca...@in...>
> I had some problem.
> My problem was caused by driver of printer. There was a series of HP drivers that were a bit buggy and by that I mean they turned
off the floating point exception handling and never returned the state.

Thanks, Daniele.

> You might want to save the state before the call, then return the state after the call.

Save the state of what? Could you explain how I would do this?

> You can try if is this problem add a new printer and set this new printer like defaut (for example print pdf).

Ok, will do. I might also try deleting all printer drivers and see if that helps

Tom
.

Re: Zero length vectors causing problems

[Daniele C. <dca...@in...>]

<html>
<body>
I had some problem.<br>
My problem was caused by driver of printer. There was a series of HP
drivers that were a bit buggy and by that I mean they turned off the
floating point exception handling and never returned the state. <br>
You might want to save the state before the call, then return the state
after the call.<br>
You can try if is this problem add a new printer and set this new printer
like defaut (for example print pdf).<br>
Daniele<br><br>
At 12.55 11/11/04, you wrote:<br>
<blockquote type=cite class=cite cite="">Hi,<br><br>
I am getting a lot of Exceptions like this:<br><br>
&quot;MyApp caused an exception of type 10H in module
Quesa.DLL...&quot;<br><br>
when running Quesa on a Windows 98 machine. I believe that they are
caused by passing zero-length vectors or zero-size polygons,<br>
which I imagine might cause Quesa to try to divide by zero.<br><br>
It is difficult to eradicate all possibility of a zero-length vector from
my program. Is there a way to get Quesa to just ignore<br>
zero-length vectors and zero-size triangles / polygons?<br><br>
If they are all divide-by-zeroes could the '/' operator be somehow
overloaded to check first?<br><br>
Tom<br><br>
<br><br>
-------------------------------------------------------<br>
This SF.Net email is sponsored by:<br>
Sybase ASE Linux Express Edition - download now for FREE<br>
LinuxWorld Reader's Choice Award Winner for best database on Linux.<br>
<a href="http://ads.osdn.com/?ad_id=5588&amp;alloc_id=12065&amp;op=click" eudora="autourl">http://ads.osdn.com/?ad_id=5588&amp;alloc_id=12065&amp;op=click</a><br>
_______________________________________________<br>
Quesa-develop mailing list<br>
Que...@li...<br>
<a href="https://lists.sourceforge.net/lists/listinfo/quesa-develop" eudora="autourl">https://lists.sourceforge.net/lists/listinfo/quesa-develop</a></blockquote></body>
</html>

Zero length vectors causing problems

[Tom S. <to...@as...>]

Hi,

I am getting a lot of Exceptions like this:

"MyApp caused an exception of type 10H in module Quesa.DLL..."

when running Quesa on a Windows 98 machine. I believe that they are caused by passing zero-length vectors or zero-size polygons,
which I imagine might cause Quesa to try to divide by zero.

It is difficult to eradicate all possibility of a zero-length vector from my program. Is there a way to get Quesa to just ignore
zero-length vectors and zero-size triangles / polygons?

If they are all divide-by-zeroes could the '/' operator be somehow overloaded to check first?

Tom

quesa.org

[Mike W. <wh...@ar...>]

Last week i changed the DNS entry for quesa.org so instead of 
pointing the Artifice hosted Quesa server it pointed to the 
Sourceforge Quesa server (IP number 66.35.250.209).

When requesting a page from quesa.org the new server returns a 404 
error, though quesa.sourceforge.net returns good pages.

The following URLs should all be equivalent if not for some unusual 
server configuration on the SourceForge side. Unfortunately only one 
of them works.

http://www.quesa.org - 404 Not Found
http://66.35.250.209 - 404 Not Found
http://quesa.sourceforge.net - Quesa homepage

It looks to me like to make this work quesa.org will have to come 
back to being an Artifice hosted server which issues a redirect to 
http://quesa.sourceforge.net. There would be no real URLs with 
quesa.org, they would all be redirected to the quesa.sourceforge.net 
domain.

Let me know if anyone else has a different idea of how to make it work.

Mike

quesa.org not working

[Joseph J. S. <jo...@st...>]

I have a user who thought that Quesa was dead, because Google 
searches turned up a bunch of links to and within www.quesa.org.  And 
that returns a 404 Not Found error.

I haven't been keeping up with the Quesa list lately, so it's 
entirely possible this was just discussed five minutes ago.  But in 
case it was not, what's the status of getting quesa.org to point to 
quesa.sourceforge.net?

Best,
- Joe

-- 
,------------------------------------------------------------------.
|    Joseph J. Strout         Check out the Mac Web Directory:     |
|    jo...@st...           http://www.macwebdir.com/            |
`------------------------------------------------------------------'

Re: hash function optimization

[James W. W. <ja...@wr...>]

Dair Grant <da...@re...> wrote:

>This was actually one of the changes that came up when looking at
>Roger's changes to the way instance data is found; the existing hash
>function turned out to be really quite suboptimal (use of % generates an
>expensive divide), and so replacing that was definitely a good idea.
>
>I actually had it changed here too, but never got round to checking it
>in - thanks! :-)

Did you change it the same way?  I didn't get rid of the % operation 
entirely, just reduced from 4 times to one.  If we could require that 
the table size is always a power of 2, then we could replace x % 
tableSize by x & (tableSize - 1).  Do you suppose that would be 
worthwhile?
-- 
   James W. Walker, ScriptPerfection Enterprises, Inc.
   <http://www.write-brain.com/>

Re: hash function optimization

[Dair G. <da...@re...>]

James W. Walker wrote:

>Just so you know that I'm not optimizing without profiling:  In a=20
>particular situation, picking with lots of complex objects, Shark=20
>showed that about 16% of the time was being spent in=20
>e3hash_find_node.  After the optimization that I just checked in,=20
>that went down to about 6%.

The perils of not posting for a while... :-)

This was actually one of the changes that came up when looking at
Roger's changes to the way instance data is found; the existing hash
function turned out to be really quite suboptimal (use of % generates an
expensive divide), and so replacing that was definitely a good idea.

I actually had it changed here too, but never got round to checking it
in - thanks! :-)


-dair
___________________________________________________
mailto:dair+refnum.com       http://www.refnum.com/

hash function optimization

[James W. W. <ja...@wr...>]

Just so you know that I'm not optimizing without profiling:  In a 
particular situation, picking with lots of complex objects, Shark 
showed that about 16% of the time was being spent in 
e3hash_find_node.  After the optimization that I just checked in, 
that went down to about 6%.
-- 
   James W. Walker, ScriptPerfection Enterprises, Inc.
   <http://www.write-brain.com/>

Re: Mach-O plug-ins for Quesa

[James W. W. <ja...@wr...>]

Peter Michelsen <Pet...@mi...> wrote:

>Looks like you where busy over the weekend:-)

Yes... not to mention giving candy to small extortionists. :-)

>I read over the CVS submit of  E3MacSystem but there is no change to
>the actual CFM version for plug-in loading. Is this right? You could
>just put a comment next to "ResolveAliasFile" that a better option
>would be "FSResolveAliasFileWithMountFlags" on OS X as there is a
>problem with "ResolveAliasFile" hanging, causing strange results (
>Actually what I was seeing and not asserting). Or use the code I
>submitted to the list with a system checking if-state to keep OS 7
>support with the value from the function that calls it "isOnOSX".

I just submitted a change using ResolveAliasFileWithMountFlags 
instead of FSResolveAliasFileWithMountFlags.  I bet that will solve 
your problem, let me know.

>Two new question?
>
>The RayShader renderer should this work like any other Quickdraw3D 
>software renderer, e.g. as a plug-in?

Right.  But expect it to be a lot slower than the interactive renderer.

>And how would you rate it's completeness as a project ( mostly done 
>or just started ) ?

I don't know that much about it.  Maybe someone else here can respond to that.

>The other question I see you are using CodeWarrior 9, did you find 
>many problems upgrading for CodeWarrior 8.3 and was it worth it?
>I ask, because we are thinking about upgrading but am unsure about 
>the effort needed in the change?

For me, the main advantage of CW 9 is better support for building 
packaged apps, frameworks, and other bundles.  If that's important to 
you, then it's worth it.  I wouldn't say there were many problems, 
once I learned to turn off the "language parser".
-- 
   James W. Walker, ScriptPerfection Enterprises, Inc.
   <http://www.write-brain.com/>

Error report on application exit

[Tom S. <to...@as...>]

Hi,

This is with Quesa 1.6d18 on Windows XP:

When my Quesa application exits, the following debug messages are sent silently:

CMD_DELAPMESSAGE escape not supported

WGL Message: WGL: DDraw: UpdateClipList: GetClipList (1): DDERROR = DDERR_GENERIC

WGL Message: WGL: DDraw: UpdateClipList: GetClipList (2): DDERROR = DDERR_GENERIC

CMD_DELAPMESSAGE escape not supported

CMD_DELAPMESSAGE escape not supported



Can anyone tell me what these mean?
Is the application doing something wrong to cause these errors?

Thanks,

Tom

Re: Mach-O plug-ins for Quesa

[Peter M. <Pet...@mi...>]

Hi James,

Looks like you where busy over the weekend:-)

I read over the CVS submit of  E3MacSystem but there is no change to 
the actual CFM version for plug-in loading. Is this right? You could 
just put a comment next to "ResolveAliasFile" that a better option 
would be "FSResolveAliasFileWithMountFlags" on OS X as there is a 
problem with "ResolveAliasFile" hanging, causing strange results ( 
Actually what I was seeing and not asserting). Or use the code I 
submitted to the list with a system checking if-state to keep OS 7 
support with the value from the function that calls it "isOnOSX".

Two new question?

The RayShader renderer should this work like any other Quickdraw3D 
software renderer, e.g. as a plug-in?
And how would you rate it's completeness as a project ( mostly done or 
just started ) ?

The other question I see you are using CodeWarrior 9, did you find many 
problems upgrading for CodeWarrior 8.3 and was it worth it?
I ask, because we are thinking about upgrading but am unsure about the 
effort needed in the change?

Thanks for any answer. Peter Michelsen.

On 1 Nov 2004, at 01:57, James W. Walker wrote:

> Following the discussion of plug-in loading last week, I decided to 
> implement Mach-O plugin loading.  I decided that to try to make 
> plug-ins cross runtime architectures (Mach-O Quesa loading CFM plugin 
> or vice versa) would be too much effort for too little benefit.  So, a 
> CFM version of Quesa still only sees CFM shared library plug-ins.  On 
> the other hand, a Mach-o framework version of Quesa looks for Mach-o 
> bundle plugins, which are recognized by the extension ".quesaplug".
>
> In order for the framework version of Quesa to load its resources, it 
> needs to have an Info.plist file specifying the bundle identifier 
> "org.Quesa.Quesa", and to specify an initialization function 
> E3MacMachoFrameworkInit.  I've added a CodeWarrior 9 project to build 
> frameworks that way.  The XCode project will also need to be updated, 
> but I'm not sure I'm sufficiently comfortable with XCode to be mucking 
> about with it.
>
> In order to test all this, I updated RayShade to build a Mach-O plugin 
> version of it.
> --
> <http://www.jwwalker.com/>
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Sybase ASE Linux Express Edition - download now for FREE
> LinuxWorld Reader's Choice Award Winner for best database on Linux.
> http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
> _______________________________________________
> Quesa-develop mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quesa-develop
>

Mach-O plug-ins for Quesa

[James W. W. <os...@jw...>]

Following the discussion of plug-in loading last week, I decided to 
implement Mach-O plugin loading.  I decided that to try to make 
plug-ins cross runtime architectures (Mach-O Quesa loading CFM plugin 
or vice versa) would be too much effort for too little benefit.  So, a 
CFM version of Quesa still only sees CFM shared library plug-ins.  On 
the other hand, a Mach-o framework version of Quesa looks for Mach-o 
bundle plugins, which are recognized by the extension ".quesaplug".

In order for the framework version of Quesa to load its resources, it 
needs to have an Info.plist file specifying the bundle identifier 
"org.Quesa.Quesa", and to specify an initialization function 
E3MacMachoFrameworkInit.  I've added a CodeWarrior 9 project to build 
frameworks that way.  The XCode project will also need to be updated, 
but I'm not sure I'm sufficiently comfortable with XCode to be mucking 
about with it.

In order to test all this, I updated RayShade to build a Mach-O plugin 
version of it.
--
<http://www.jwwalker.com/>

Re: quesa.org still wrong

[Kevin M. <mat...@ar...>]

I thought this had been taken care of long since.  I'll look into it myself later today.  (Was it ever switched over at all?  Could there be a conf file version control issue at our server?  Or it's just a pure service drop-out...)

Thanks,

Kevin


On Sun, 31 Oct 2004 09:29:01 -0800, James W. Walker wrote:
> quesa.org still does not point to quesa.sourceforge.net, as we 
> discussed several months ago.  What's the holdup?
> --
> <http://www.jwwalker.com/>
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by:
> Sybase ASE Linux Express Edition - download now for FREE
> LinuxWorld Reader's Choice Award Winner for best database on Linux.
> http://ads.osdn.com/?ad_id=5588&alloc_id=12065&op=click
> _______________________________________________
> Quesa-develop mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quesa-develop
> 

quesa.org still wrong

[James W. W. <os...@jw...>]

quesa.org still does not point to quesa.sourceforge.net, as we 
discussed several months ago.  What's the holdup?
--
<http://www.jwwalker.com/>