quesa-develop — Mailing list for Quesa development

Re: MS VC++ 6 issues

[Jose' C. <cru...@ce...>]

Il giorno 18/gen/05, alle 02:34, James W. Walker ha scritto:

> In the original code, it was the TOCEntry structure, not its 
> CQ3ObjectRef member, that was being initialized with an initializer 
> list.  The TOCEntry structure is an aggregate, being a structure that 
> does not have constructors, private or protected members, base 
> classes, or virtual functions.  Therefore, the text you quote here 
> does not explain why one should not be able to initialize TOCEntry 
> with an initializer list.

the error is that, so probably there's a implicit constructor since 
there's a constructor in their members... boh..

>
> I've heard that VC++ 7 is much more standards-compliant.  I wonder why 
> you don't upgrade?  Is it a matter of cost, or are you in a situation 
> where it is considered too risky to change compilers?
>

really I use VC 6.0 only for Quesa (in production I use CW 8.3 because 
the cross compiling abilities) but seems to me that is a reasonable 
minimum target, given the huge installed base


Pax et Bonum

# dott. Jose' Cruanyes Aguilar  -  C.E. Soft srl
#  Pzza. Firenze,4 MILANO  -  XX Settembre 10, CREMONA
#   02,33603122                0372,460602

Re: MS VC++ 6 issues

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

On Jan 17, 2005, at 3:19 PM, Jose' Cruanyes wrote:

> the error is C2552
>
> from the VC 6.0 manual
> ------------------
> Compiler Error C2552
>
> 'identifier' : non-aggregates cannot be initialized with initializer=20=

> list
>
> The specified identifier was incorrectly initialized.
>
> An initializer list is needed to initialize the following types:
> 	=95 	 An array
>
> 	=95 	A class, structure, or union that does not have =
constructors,=20
> private or protected members, base classes, or virtual functions
>
> These types are known as =93aggregates.=94
> ------------------
>
> with these criteria the error is legitimate since the CQ3ObjectRef has=20=

> a constructor
> now the question is 'are these criteria legitimate?'

In the original code, it was the TOCEntry structure, not its=20
CQ3ObjectRef member, that was being initialized with an initializer=20
list.  The TOCEntry structure is an aggregate, being a structure that=20
does not have constructors, private or protected members, base classes,=20=

or virtual functions.  Therefore, the text you quote here does not=20
explain why one should not be able to initialize TOCEntry with an=20
initializer list.

I've heard that VC++ 7 is much more standards-compliant.  I wonder why=20=

you don't upgrade?  Is it a matter of cost, or are you in a situation=20
where it is considered too risky to change compilers?
--
<http://www.jwwalker.com/>

Re: MS VC++ 6 issues

[Jose' C. <cru...@ce...>]

Il giorno 17/gen/05, alle 21:23, James W. Walker ha scritto:

> I guess I can get used to not using the and, or, and not keywords,=20
> maybe they're not standard.

probably we can do it with a #define


>   Making a class a friend to itself is bizarre, but maybe it won't=20
> come up too often.

I don't know about standards in this case, but it can have a sense, the=20=

error given because in a method is used a private (if changed to=20
protected it works) member of another object of the same class

coming from Microsoft pushing a bit the 'private property' concept is=20
no surprising :-)


>   I'm really worried about the change below.  What was the compiler=20
> complaining about?  Looks like standard C++ to me, and certainly seems=20=

> to be permitted according to sections 8.5.1 and 12.6.1 of the 1998 C++=20=

> standard.

the error is C2552

from the VC 6.0 manual
------------------
Compiler Error C2552

'identifier' : non-aggregates cannot be initialized with initializer=20
list

The specified identifier was incorrectly initialized.

An initializer list is needed to initialize the following types:
	=95 	 An array

	=95 	A class, structure, or union that does not have =
constructors,=20
private or protected members, base classes, or virtual functions

These types are known as =93aggregates.=94
------------------

with these criteria the error is legitimate since the CQ3ObjectRef has=20=

a constructor
now the question is 'are these criteria legitimate?'



Pax et Bonum

# dott. Jose' Cruanyes Aguilar  -  C.E. Soft srl
#  Pzza. Firenze,4 MILANO  -  XX Settembre 10, CREMONA
#   02,33603122                0372,460602

Re: MS VC++ 6 issues

[Roger H. <rog...@mi...>]

I don't know what  the actual message from the compiler was, but maybe 
it was tying itself in knots
thinking that the refID field was being set to itself rather than a 
variable of the same name.
Just a hunch, I don't use the windows C++ compiler.

Making a class its own friend seems pointless, and as you say, bizarre. 
Anyway that class will be
merged into OpaqueTQ3Object in a couple of weeks if all goes well.

Roger.



On Monday, January 17, 2005, at 08:23  pm, James W. Walker wrote:

> I guess I can get used to not using the and, or, and not keywords, 
> maybe they're not standard.  Making a class a friend to itself is 
> bizarre, but maybe it won't come up too often.  I'm really worried 
> about the change below.  What was the compiler complaining about?  
> Looks like standard C++ to me, and certainly seems to be permitted 
> according to sections 8.5.1 and 12.6.1 of the 1998 C++ standard.
>
>> *** 939,947 ****
>>   					if (labelIter != instanceData->mLabelMap->end())
>>   					{
>> ! 						TOCEntry	tocEntry = {
>> ! 							refID,
>> ! 							labelIter->second,
>> ! 							CQ3ObjectRef()
>> ! 						};
>>   						instanceData->mTOC->push_back( tocEntry );
>>   					}
>> --- 939,948 ----
>>   					if (labelIter != instanceData->mLabelMap->end())
>>   					{
>> ! 						TOCEntry	tocEntry;
>> ! 						
>> ! 						tocEntry.refID = refID;
>> ! 						tocEntry.objLocation = labelIter->second;
>> ! 						tocEntry.object = CQ3ObjectRef();
>> ! 						
>
> --
> <http://www.jwwalker.com/>
>
>
>
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Quesa-develop mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quesa-develop
>

MS VC++ 6 issues

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

I guess I can get used to not using the and, or, and not keywords, 
maybe they're not standard.  Making a class a friend to itself is 
bizarre, but maybe it won't come up too often.  I'm really worried 
about the change below.  What was the compiler complaining about?  
Looks like standard C++ to me, and certainly seems to be permitted 
according to sections 8.5.1 and 12.6.1 of the 1998 C++ standard.

> *** 939,947 ****
>   					if (labelIter != instanceData->mLabelMap->end())
>   					{
> ! 						TOCEntry	tocEntry = {
> ! 							refID,
> ! 							labelIter->second,
> ! 							CQ3ObjectRef()
> ! 						};
>   						instanceData->mTOC->push_back( tocEntry );
>   					}
> --- 939,948 ----
>   					if (labelIter != instanceData->mLabelMap->end())
>   					{
> ! 						TOCEntry	tocEntry;
> ! 						
> ! 						tocEntry.refID = refID;
> ! 						tocEntry.objLocation = labelIter->second;
> ! 						tocEntry.object = CQ3ObjectRef();
> ! 						

--
<http://www.jwwalker.com/>

E3Torus_SetRatio does not do Q3Shared_Edited. A bug ?

[Roger H. <rog...@mi...>]

Nearly all the other 'Set' routines contain a call to Q3Shared_Edited.

For some reason E3Torus_SetRatio does not. I will presume this is
a bug unless someone tells me otherwise.

Roger.

Re: texture UV animation problems

[Keith W. <kw...@cs...>]

On Thu, 13 Jan 2005, James W. Walker wrote:

> On Jan 12, 2005, at 11:12 PM, Keith Wiley wrote:
>
>> Does anyone know anything about this?  What is a reliable way to get 
>> textures to animate?  Apparantly there are some detrimental bugs 
>> associated with UV transforming.
>
> I don't have experience with either alpha textures or real-time varying UV 
> transforms, and I don't see anything in the bug tracker that sounds like 
> this.  If you could cook up a self-contained example and include that with a 
> bug report, it would be more likely to be fixed.

Yeah, I need to learn how to submit bugs, I've found a few.

> Have you tried this with more than one graphics card, or with a software 
> OpenGL renderer?  This is probably a Quesa bug, but it's conceivable that a 
> graphics card bug could be involved.

It's doing it on multiple machines which have different cards.

Thanks.

________________________________________________________________________
Keith Wiley         kw...@cs...         http://www.unm.edu/~keithw

"Yet mark his perfect self-contentment, and hence learn his lesson,
that to be self-contented is to be vile and ignorant, and that to
aspire is better than to be blindly and impotently happy."
                                            --  Edwin A. Abbott, Flatland
________________________________________________________________________

Re: texture UV animation problems

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

On Jan 12, 2005, at 11:12 PM, Keith Wiley wrote:

> Does anyone know anything about this?  What is a reliable way to get 
> textures to animate?  Apparantly there are some detrimental bugs 
> associated with UV transforming.

I don't have experience with either alpha textures or real-time varying 
UV transforms, and I don't see anything in the bug tracker that sounds 
like this.  If you could cook up a self-contained example and include 
that with a bug report, it would be more likely to be fixed.

Have you tried this with more than one graphics card, or with a 
software OpenGL renderer?  This is probably a Quesa bug, but it's 
conceivable that a graphics card bug could be involved.
--
<http://www.jwwalker.com/>

Re: C++ changes

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

On Jan 13, 2005, at 7:14 AM, Roger Holmes wrote:

> Looking at the existing code of E3Box_SetFaceAttributeSet, it
> only ever returns kQ3Failure. Can I presume this is an error?

Sure looks like an error to me.
--
<http://www.jwwalker.com/>

Re: C++ changes

[Roger H. <rog...@mi...>]

Looking at the existing code of E3Box_SetFaceAttributeSet, it
only ever returns kQ3Failure. Can I presume this is an error?

Roger.

Re: texture UV followup

[Keith W. <kw...@cs...>]

I have also noticed that this problem is connected to the camera angle 
of all things.  When this problem occurs (as stated, for alpha textures 
primarily) the problem occurs in most camera angles, but as I move 
around and change the camera angle, every once in a while there is an 
eccentric camera angle where the animation suddently starts happening.  
Once I switch out of that camera angle it stops immediately.  If I go 
back to that camera angle, it works again.

Isn't that strange?  What the heck is going on?

I would really like to get this working.  It would be useful for a nice 
water effect.

If anyone can help, I would appreciate it.  Thanks again.

Cheers!


________________________________________________________________________
Keith Wiley         kw...@cs...         http://www.unm.edu/~keithw

"Yet mark his perfect self-contentment, and hence learn his lesson,
that to be self-contented is to be vile and ignorant, and that to
aspire is better than to be blindly and impotently happy."
                                           --  Edwin A. Abbott, Flatland
________________________________________________________________________

texture UV animation problems

[Keith W. <kw...@cs...>]

In Meshwork I have created and exported an object.  In my program I use 
pretty simply UV transformation to make the texture slide over the 
surface of the object.  Now, if in Meshwork I check the button for 
opaque or transparent, the animation works perfectly, but if I check the 
button for alpha, the program exhibits no animation.  The code is 
identical in these cases, only the object is being altered (and only its 
texture assignment, not the geometry).

Does anyone know anything about this?  What is a reliable way to get 
textures to animate?  Apparantly there are some detrimental bugs 
associated with UV transforming.

I have attempt to manually animate the UV coordinates by simply 
"pushing" the UV assignment of every vertex in the object at regular 
intervals.  This works perfectly at first, but then flips around starts 
sliding backwards periodically.  I think this occurs when the UV 
coordinates wrap around the end of the texture.

Any thoughts on this?

Thanks.


________________________________________________________________________
Keith Wiley         kw...@cs...         http://www.unm.edu/~keithw

"Yet mark his perfect self-contentment, and hence learn his lesson,
that to be self-contented is to be vile and ignorant, and that to
aspire is better than to be blindly and impotently happy."
                                           --  Edwin A. Abbott, Flatland
________________________________________________________________________

3D rotation question

[Keith W. <kw...@cs...>]

I hope someone can help with this.  I have a 3D object which is a single 
flat square, two triangles.  The square is initially positioned 
"horizontal", parallel to the ground plane.  The X and Z axes are aligned 
with the axes of the square incidently.

I want to alter the orientation by first pitching the square some, so it 
becomes a ramp with two sides, and then rotating the ramp around a 
vertical axis so I can face the ramp anyway I want (or alternatively by 
rotating the square in some direction and then pitch the square into a 
ramp in the new direction).  The problem of course is that I can't yaw and 
then pitch or pitch and then yaw, because in either case, the first 
rotation changes the axes so the second rotation doesn't do what I 
want it to do.  How do I maintain or regain the original axes?

Can I pitch the square and then regain the vertical axis so I can yaw the 
ramp to face any direction I want?  How do I do this?

Thanks.

________________________________________________________________________
Keith Wiley         kw...@cs...         http://www.unm.edu/~keithw

"Yet mark his perfect self-contentment, and hence learn his lesson,
that to be self-contented is to be vile and ignorant, and that to
aspire is better than to be blindly and impotently happy."
                                            --  Edwin A. Abbott, Flatland
________________________________________________________________________

Re: CVS fault?

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

>Have now checked in fixes to E3Memory.h and .c to fix it.

Good, but there is still a problem:  When a leak is detected, 
E3Memory_DumpRecording crashes.
-- 
   James W. Walker, ScriptPerfection Enterprises, Inc.
   <http://www.write-brain.com/>

Re: CVS fault?

[Roger H. <rog...@mi...>]

On Tuesday, January 11, 2005, at 06:57  pm, James W. Walker wrote:

> I got an access violation right away, trying to use it with Geom Test. 
>  This is in debug code, so if you did not build it with Q3_DEBUG, you 
> might not get the crash.
>
> The problem is where e3root_new uses the macros PREVLINK and NEXTLINK, 
> which use E3Class_FindInstanceData.  The object being passed to 
> E3Class_FindInstanceData does not seem to be valid.
>

Have now checked in fixes to E3Memory.h and .c to fix it.

Roger.

Re: CVS fault?

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

>I have now committed the first batch of changes, the changes should 
>be consistent and it should still work, if not then please let me 
>know how it fails.
>It works fine with 3D World.

I got an access violation right away, trying to use it with Geom 
Test.  This is in debug code, so if you did not build it with 
Q3_DEBUG, you might not get the crash.

The problem is where e3root_new uses the macros PREVLINK and 
NEXTLINK, which use E3Class_FindInstanceData.  The object being 
passed to E3Class_FindInstanceData does not seem to be valid.
-- 
   James W. Walker, ScriptPerfection Enterprises, Inc.
   <http://www.write-brain.com/>

Re: CVS fault?

[Roger H. <rog...@mi...>]

On Tuesday, January 11, 2005, at 05:23  pm, James W. Walker wrote:

> CodeWarrior 9 has std::string::back(), but I see that it's not in the 
> 1998 C++ standard.  I have removed the calls to std::string::back() 
> and verified that it now works with CodeWarrior 8.

Thanks. It now compiles with CW 8.3

I have now committed the first batch of changes, the changes should be 
consistent and it should still work, if not then please let me know how 
it fails.
It works fine with 3D World.

Roger.

Re: CVS fault?

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

On Jan 11, 2005, at 5:49 AM, Roger Holmes wrote:

> There seems to be a small fault with the current CVS source.
>
> In EE3FFR_3DMF_text.c is a routine e3fformat_3dmf_text_read_toc.
>
> The third line reads:
>
> 	if ( (not tocLabel.empty()) && (tocLabel.back() == '>'))
>
> Code Warrior 8.3 gives the message "undefined identifier 'back' "
>
> and then six more errors which may be consequential.

Sorry about that.  CodeWarrior 9 has std::string::back(), but I see 
that it's not in the 1998 C++ standard.  I have removed the calls to 
std::string::back() and verified that it now works with CodeWarrior 8.
--
<http://www.jwwalker.com/>

CVS fault?

[Roger H. <rog...@mi...>]

There seems to be a small fault with the current CVS source.

In EE3FFR_3DMF_text.c is a routine e3fformat_3dmf_text_read_toc.

The third line reads:

	if ( (not tocLabel.empty()) && (tocLabel.back() == '>'))

Code Warrior 8.3 gives the message "undefined identifier 'back' "

and then six more errors which may be consequential.



I will delay checking in my stuff until this is looked at.



Many thanks,

Roger.

Re: C++ changes

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

>Changed version of E3ClassTree now complete, shall I check it in or 
>would you like a chance to see it first?

I'd be interested in seeing it.
-- 
   James W. Walker, ScriptPerfection Enterprises, Inc.
   <http://www.write-brain.com/>

Re: C++ changes

[Roger H. <rog...@mi...>]

Changed version of E3ClassTree now complete, shall I check it in or 
would you like a chance to see it first?

Re: C++ changes

[Roger H. <rog...@mi...>]

Ok, standing back for a moment I agree.

I think all the static methods should stay in E3ClassTree whilst the 
non static ones
should go into OpaqueTQ3Object. Of course E3ClassTree will have to be a 
friend
of OpaqueTQ3Object. E3ClassTree become a stand alone class with no data 
in it.

Its quite a small change but it should make things clearer.



On Friday, January 7, 2005, at 06:17  pm, James W. Walker wrote:

> Roger Holmes <rog...@mi...> wrote:
>
>> I have found a good solution, the real root class is now E3ClassTree, 
>> and
>> OpaqueTQ3Object inherits from it (and currently adds no fields or 
>> methods).
>
> I don't get it.  This would let OpaqueTQ3Object methods access members 
> of E3ClassTree, but not the other way around, which is what you want.
> -- 
>   James W. Walker, ScriptPerfection Enterprises, Inc.
>   <http://www.write-brain.com/>
>
>
> -------------------------------------------------------
> The SF.Net email is sponsored by: Beat the post-holiday blues
> Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek.
> It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt
> _______________________________________________
> Quesa-develop mailing list
> Que...@li...
> https://lists.sourceforge.net/lists/listinfo/quesa-develop
>

Re: C++ changes

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

Roger Holmes <rog...@mi...> wrote:

>I have found a good solution, the real root class is now E3ClassTree, and
>OpaqueTQ3Object inherits from it (and currently adds no fields or methods).

I don't get it.  This would let OpaqueTQ3Object methods access 
members of E3ClassTree, but not the other way around, which is what 
you want.
-- 
   James W. Walker, ScriptPerfection Enterprises, Inc.
   <http://www.write-brain.com/>

Re: C++ changes

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

On Jan 7, 2005, at 5:04 AM, Roger Holmes wrote:

> I have found a good solution, the real root class is now E3ClassTree, 
> and
> OpaqueTQ3Object inherits from it (and currently adds no fields or 
> methods).

So, an object is a class tree?  Doesn't make sense to me.
--
<http://www.jwwalker.com/>

Re: C++ changes

[Roger H. <rog...@mi...>]

> But why you just doesn't declares a new routine that takes the class 
> size instead of the instance size, and add a flag to note what kind of 
> class is it?

Because the extra parameter is only temporary, and in any case I do not 
want to
leave the CVS source in an inconsistent state for a long period whilst 
I check in the
changes to every file to add the extra parameter, and then again when I 
remove it later.



>> 1) Should every E3ClassTree routine be a friend or a method?
>> 2) If a method, then should ones with a TQ3Object take it as their 
>> 'this' ?
>
> I second the James proposal of a class for tree methods

I have found a good solution, the real root class is now E3ClassTree, 
and
OpaqueTQ3Object inherits from it (and currently adds no fields or 
methods).
E3ClassTree currently has lots of friend routines, but these will soon 
be replaced
with methods, though the friend routines will temporarily remain until 
all calls to
them can be changed, but they will just be inline calls to the real 
method anyway.
Hopefully I can get rid of them in a week or two.




>> 3) If so, should the method names be as in option A, B or C ?
>
> IMHO Always whatever makes sense (in your case option C)


Great.


I hope to check in E3ClassTree  on Monday, or later today if there are 
no snags to iron out.