| | | | |
In lugnet.cad.dev, Don Heyse wrote:
|
Kevin, if you want to test it I put a beta of just the executable
here.
|
Apparently this is working for others, but when I run it (on an original MacBook
in OS 10.4), all I get is a bus error. Did you compile it as a Leopard app?
--Travis
| | | | | | | | | | | | | In lugnet.cad.dev, Travis Cobbs wrote:
|
In lugnet.cad.dev, Don Heyse wrote:
|
Kevin, if you want to test it I put a beta of just the executable
here.
|
Apparently this is working for others, but when I run it (on an original
MacBook in OS 10.4), all I get is a bus error. Did you compile it as a
Leopard app?
|
Err, I rushed it and sorta botched the build. So a few files got built
with no PPC code. I rebuilt it, but didnt get a chance to borrow an
older machine to test it on yet.
Hey, maybe (probably) you know this. Whats involved in supporting
a G3, or say OS 10.3? I dont know my way around the compiler switches
well enough yet to turn off altivec or use the older libraries.
Don
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
|
Err, I rushed it and sorta botched the build. So a few files got built
with no PPC code. I rebuilt it, but didnt get a chance to borrow an
older machine to test it on yet.
|
Actually, I dont have a PPC Mac, just the previous OS version. I believe that
if you dont get everything right for the universal stuff, it will give you an
error at link time. I know thats what it does to me if any of the libraries
arent universal and I try to do a universal build.
|
Hey, maybe (probably) you know this. Whats involved in supporting
a G3, or say OS 10.3? I dont know my way around the compiler switches
well enough yet to turn off altivec or use the older libraries.
|
No idea. I know how to set up the OS setting in XCode, but I dont know
off-hand even how to do that in a Makefile.
--Travis
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Travis Cobbs wrote:
|
In lugnet.cad.dev, Don Heyse wrote:
|
Err, I rushed it and sorta botched the build. So a few files got built
with no PPC code. I rebuilt it, but didnt get a chance to borrow an
older machine to test it on yet.
|
Actually, I dont have a PPC Mac, just the previous OS version.
|
Thats interesting. I built it on a new laptop with the version of XCode
that comes with leopard. However I just added the arch targets to the
old makefile (but missed one of the libraries). So I guess the default
is to target leopard only. I suppose I should try to build from an
XCode project and see if its easier to target older OS versions and
CPUs with that.
|
I believe
that if you dont get everything right for the universal stuff, it will
give you an error at link time. I know thats what it does to me if any of
the libraries arent universal and I try to do a universal build.
|
It let me link in an X86 only lib without an error. Maybe there was a
warning, I didnt pay too much attention. I can imagine reasons why
you *might* want a special lib thats only called from the X86 code.
|
|
Hey, maybe (probably) you know this. Whats involved in supporting
a G3, or say OS 10.3? I dont know my way around the compiler switches
well enough yet to turn off altivec or use the older libraries.
|
No idea. I know how to set up the OS setting in XCode, but I dont know
off-hand even how to do that in a Makefile.
|
I found some documentation, but I printed it and dont have it on hand
right now. It sure looks easier to do from the XCode gui though.
So whats the oldest, junkiest Mac thats known to run the LDView executable?
Don
| | | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
|
So whats the oldest, junkiest Mac thats known to run the LDView executable?
|
No idea. Whatever it is, it needs Tiger, though, since LDView is built to
require Tiger. It presumably needs a G4.
--Travis
| | | | | | | | | | | | | | | | | | | | | In lugnet.cad.dev, Travis Cobbs wrote:
|
In lugnet.cad.dev, Don Heyse wrote:
|
So whats the oldest, junkiest Mac thats known to run the LDView
executable?
|
No idea. Whatever it is, it needs Tiger, though, since LDView is built to
require Tiger. It presumably needs a G4.
|
For what its worth, Tiger will run on a 300 MHz G3. While it certainly wasnt
fast, I dont recall ever encountering any software that simply said I need a
G4. Thats not to say theres no reason why such software couldnt or shouldnt
exist, of course.
Jim
| | | | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
|
In lugnet.cad.dev, Travis Cobbs wrote:
|
In lugnet.cad.dev, Don Heyse wrote:
|
Err, I rushed it and sorta botched the build. So a few files got built
with no PPC code. I rebuilt it, but didnt get a chance to borrow an
older machine to test it on yet.
|
Actually, I dont have a PPC Mac, just the previous OS version.
|
Thats interesting. I built it on a new laptop with the version of XCode
that comes with leopard. However I just added the arch targets to the
old makefile (but missed one of the libraries). So I guess the default
is to target leopard only. I suppose I should try to build from an
XCode project and see if its easier to target older OS versions and
CPUs with that.
|
I believe
that if you dont get everything right for the universal stuff, it will
give you an error at link time. I know thats what it does to me if any of
the libraries arent universal and I try to do a universal build.
|
It let me link in an X86 only lib without an error. Maybe there was a
warning, I didnt pay too much attention. I can imagine reasons why
you *might* want a special lib thats only called from the X86 code.
|
|
Hey, maybe (probably) you know this. Whats involved in supporting
a G3, or say OS 10.3? I dont know my way around the compiler switches
well enough yet to turn off altivec or use the older libraries.
|
No idea. I know how to set up the OS setting in XCode, but I dont know
off-hand even how to do that in a Makefile.
|
I found some documentation, but I printed it and dont have it on hand
right now. It sure looks easier to do from the XCode gui though.
|
Since I already lost this once, Im gonna stash it here just in case.
To compile something that runs on Tiger from a makefile on Leopard add
this:
CFLAGS+= -mmacosx-version-min=10.4 -isysroot /Developer/SDKs/MacOSX10.4u.sdk
LDFLAGS+= -isysroot /Developer/SDKs/MacOSX10.4u.sdk
That should make it build and run with the older Tiger sdk and libs.
Add -arch i386 -arch ppc and I should be able to make a universal
binary for 10.4 and up. (I hope)
Don
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
|
Since I already lost this once, Im gonna stash it here just in case.
To compile something that runs on Tiger from a makefile on Leopard add
this:
|
Good idea.
|
CFLAGS+= -mmacosx-version-min=10.4 -isysroot /Developer/SDKs/MacOSX10.4u.sdk
LDFLAGS+= -isysroot /Developer/SDKs/MacOSX10.4u.sdk
That should make it build and run with the older Tiger sdk and libs.
Add -arch i386 -arch ppc and I should be able to make a universal
binary for 10.4 and up. (I hope)
|
Ill be happy to be your Tiger test subject, but I can only test Intel.
--Travis
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Travis Cobbs wrote:
|
In lugnet.cad.dev, Don Heyse wrote:
|
Since I already lost this once, Im gonna stash it here just in case.
To compile something that runs on Tiger from a makefile on Leopard add
this:
|
Good idea.
|
CFLAGS+= -mmacosx-version-min=10.4 -isysroot /Developer/SDKs/MacOSX10.4u.sdk
LDFLAGS+= -isysroot /Developer/SDKs/MacOSX10.4u.sdk
That should make it build and run with the older Tiger sdk and libs.
Add -arch i386 -arch ppc and I should be able to make a universal
binary for 10.4 and up. (I hope)
|
Ill be happy to be your Tiger test subject, but I can only test Intel.
|
Ok, give this one a try.
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
That one works for LPub. The buffer isnt getting flushed to the screen right.
I have to resize or force a redraw in order. Are you calling CGLFlushDrawable
at the end of your drawing code?
--Travis
| | | | | | | | | | | | | | | | |
| |
| In lugnet.cad.dev, Travis Cobbs wrote:
> In lugnet.cad.dev, Don Heyse wrote:
> > Ok, give http://ldglite.sf.net/ldgliteosx1_2_1beta2.zip a try.
>
> That one works for LPub. The buffer isn't getting flushed to the screen
> right. I have to resize or force a redraw in order. Are you calling
> CGLFlushDrawable at the end of your drawing code?
No, I use glfinish(). I've had all sorts of problems over the years
with Apple because I often render in the front buffer to emulate
the various ancient ldlite immediate rendering modes. Does it update
ok when you spin the model with the mouse? I think that's pure back
buffer rendering (at least until you're done spinning). What about using
-n4 on the command line? That's the ldlite command to render the whole
thing before displaying it. It looks like I forced that mode when the gl
render string contains "RADEON 7000", so I must have had problems with
that one too. Really, at this point I should probably give up rendering
in the front buffer except perhaps for editing mode and maybe for the
slow software renderers.
What does ldglite print for GL_VENDOR and GL_RENDERER?
Don
| | | | | | | | | | | | | | | | |
| |
| In lugnet.cad.dev, Don Heyse wrote:
> In lugnet.cad.dev, Travis Cobbs wrote:
> > In lugnet.cad.dev, Don Heyse wrote:
> > > Ok, give http://ldglite.sf.net/ldgliteosx1_2_1beta2.zip a try.
> >
> > That one works for LPub. The buffer isn't getting flushed to the screen
> > right. I have to resize or force a redraw in order. Are you calling
> > CGLFlushDrawable at the end of your drawing code?
>
> No, I use glfinish(). I've had all sorts of problems over the years
> with Apple because I often render in the front buffer to emulate
> the various ancient ldlite immediate rendering modes. Does it update
> ok when you spin the model with the mouse? I think that's pure back
> buffer rendering (at least until you're done spinning). What about using
> -n4 on the command line? That's the ldlite command to render the whole
> thing before displaying it. It looks like I forced that mode when the gl
> render string contains "RADEON 7000", so I must have had problems with
> that one too. Really, at this point I should probably give up rendering
> in the front buffer except perhaps for editing mode and maybe for the
> slow software renderers.
>
> What does ldglite print for GL_VENDOR and GL_RENDERER?
Sorry about the self reply, but now I'm really curious about this.
I found myself an old G4 EMac with a Geforce2 and OSX Tiger for testing
and everything worked ok. It also worked on an Intel Macbook with
the Intel GMA X3100 graphics. So I guess that means I've got the
universal binary build process down.
Now I just need to fix the display bugs on some systems. I'm guessing
they use ATI chips, but it'd be nice to know for sure. It'd also be nice
to know if the -n4 arg makes it better, because if so I'll make that the default
moving forward.
Don
| | | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
> Sorry about the self reply, but now I'm really curious about this.
> I found myself an old G4 EMac with a Geforce2 and OSX Tiger for testing
> and everything worked ok. It also worked on an Intel Macbook with
> the Intel GMA X3100 graphics. So I guess that means I've got the
> universal binary build process down.
Sorry for not responding, but I didn't remember to look at it while I was at
home. I'll try to do so tonight.
> Now I just need to fix the display bugs on some systems. I'm guessing
> they use ATI chips, but it'd be nice to know for sure. It'd also be nice
> to know if the -n4 arg makes it better, because if so I'll make that the default
> moving forward.
Actually, my Intel MacBook (one of the original ones) has Intel motherboard
graphics. Probably a different model than yours, but Intel nonetheless.
--Travis
| | | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
> Sorry about the self reply, but now I'm really curious about this.
> I found myself an old G4 EMac with a Geforce2 and OSX Tiger for testing
> and everything worked ok. It also worked on an Intel Macbook with
> the Intel GMA X3100 graphics. So I guess that means I've got the
> universal binary build process down.
Here's my info:
Vendor: Intel Inc.
Renderer: Intel GMA 950 OpenGL Engine
Version: 1.2 APPLE-1.4.56
--Travis
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Travis Cobbs wrote:
> In lugnet.cad.dev, Don Heyse wrote:
> > Sorry about the self reply, but now I'm really curious about this.
> > I found myself an old G4 EMac with a Geforce2 and OSX Tiger for testing
> > and everything worked ok. It also worked on an Intel Macbook with
> > the Intel GMA X3100 graphics. So I guess that means I've got the
> > universal binary build process down.
>
> Here's my info:
>
> Vendor: Intel Inc.
> Renderer: Intel GMA 950 OpenGL Engine
> Version: 1.2 APPLE-1.4.56
Hmmm, I guess that's slightly different from the GMA X3100, which
seemed to work for me. If you get a chance, try and run it with
-n4 on the command line. That should render everything in the back
buffer like 99.99% of the opengl programs out there.
Thanks again,
Don
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
> Hmmm, I guess that's slightly different from the GMA X3100, which
> seemed to work for me. If you get a chance, try and run it with
> -n4 on the command line. That should render everything in the back
> buffer like 99.99% of the opengl programs out there.
The -n4 didn't have any noticeable effect. What exactly does n4 mean. I know
it means "draw to back buffer", but what does the n and the 4 mean?
--Travis
| | | | | | | | | | | | | | | | |
| |
| In lugnet.cad.dev, Travis Cobbs wrote:
> In lugnet.cad.dev, Don Heyse wrote:
> > Hmmm, I guess that's slightly different from the GMA X3100, which
> > seemed to work for me. If you get a chance, try and run it with
> > -n4 on the command line. That should render everything in the back
> > buffer like 99.99% of the opengl programs out there.
>
> The -n4 didn't have any noticeable effect.
Arrgh, that's not good. It worked for me whenever I had a problem
like that. I guess I'll have to use the OSX flush function. Are
you by chance using partially transparent windows, or something
fancy like that?
> What exactly does n4 mean. I know it means "draw to back buffer",
> but what does the n and the 4 mean?
It's hooked to an ldlite variable that tracks what level is
being rendered (P=1, Part=2, Model=4, Other=8). Don't ask me what
other means. I think this was documented with the 2.4 release notes
for ldlite, but I lost track of that. I posted the ldlite 2.3
release notes as the ldlite sourceforge web page at some point, but
that doesn't mention it. So I suppose it's possible that I made the
whole thing up myself 6 years ago, but my memory from that time is
gone now.
Have fun,
Don
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
> In lugnet.cad.dev, Travis Cobbs wrote:
> > The -n4 didn't have any noticeable effect.
>
> Arrgh, that's not good. It worked for me whenever I had a problem
> like that. I guess I'll have to use the OSX flush function. Are
> you by chance using partially transparent windows, or something
> fancy like that?
Nope. It starts out partially under my dock, which I keep on the left. I can't
remember if I ever moved it completely out from under the dock or not, but I'm
pretty sure I did. (As a note, I looked around, and it appears that
HIWindowGetAvailablePositioningBounds is the Carbon API to use to avoid
overlapping the Dock; GLUT REALLY should use that automatically when you ask it
for a maximized window.)
--Travis
| | | | | | | | | | | | | | | | |
| |
| In lugnet.cad.dev, Travis Cobbs wrote:
> In lugnet.cad.dev, Don Heyse wrote:
> > In lugnet.cad.dev, Travis Cobbs wrote:
> > > The -n4 didn't have any noticeable effect.
> >
> > Arrgh, that's not good. It worked for me whenever I had a problem
> > like that. I guess I'll have to use the OSX flush function. Are
> > you by chance using partially transparent windows, or something
> > fancy like that?
>
> Nope. It starts out partially under my dock, which I keep on the left.
> I can't remember if I ever moved it completely out from under the dock
> or not, but I'm pretty sure I did. (As a note, I looked around, and it
> appears that HIWindowGetAvailablePositioningBounds is the Carbon API
> to use to avoid overlapping the Dock; GLUT REALLY should use that
> automatically when you ask it for a maximized window.)
Hmmm, a partially obscured window. Sounds like a clue. I'll have to
run some tests with a side mounted dock on the G4. And I should probably
look back and see what I did on windows when the annoying popup tooltips
were driving me nuts.
Thanks,
Don
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
> Hmmm, a partially obscured window. Sounds like a clue. I'll have to
> run some tests with a side mounted dock on the G4. And I should probably
> look back and see what I did on windows when the annoying popup tooltips
> were driving me nuts.
Well, I pulled the window out from under the dock, and that didn't help. On an
unrelated note, why will it only open models in the current directory? If I
give the full path to a model in a different directory, it fails to load it.
--Travis
| | | | | | | | | | | | | | | | |
| |
| In lugnet.cad.dev, Travis Cobbs wrote:
> In lugnet.cad.dev, Don Heyse wrote:
> > Hmmm, a partially obscured window. Sounds like a clue. I'll have to
> > run some tests with a side mounted dock on the G4. And I should probably
> > look back and see what I did on windows when the annoying popup tooltips
> > were driving me nuts.
>
> Well, I pulled the window out from under the dock, and that didn't help.
Yeah, but the hint helped me. With the side mounted dock I was able
to reproduce the problem (or a problem at least) on the G4. I cleaned up
some of the buffer swapping and window reshaping code, and it seems
to work better for me now. Of course I seem to be moving further and
further away from glut with every fix. Oh well. If you want, you could
try out this one and see if it's better on your system.
http://ldglite.sf.net/ldgliteosx1_2_3beta5.zip
> On an unrelated note, why will it only open models in the current
> directory? If I give the full path to a model in a different directory,
> it fails to load it.
I've started looking into that. It does the same thing on linux, I think.
I suspect I broke that a while ago when I added some code on Windows to
find the path to the executable in order to run lddesignpad plugins.
It's been a long time since I did much testing on anything but windows.
Have fun,
Don
| | | | | | | | | | | | | | | | | In lugnet.cad.dev, Don Heyse wrote:
> Yeah, but the hint helped me. With the side mounted dock I was able
> to reproduce the problem (or a problem at least) on the G4. I cleaned up
> some of the buffer swapping and window reshaping code, and it seems
> to work better for me now. Of course I seem to be moving further and
> further away from glut with every fix. Oh well. If you want, you could
> try out this one and see if it's better on your system.
>
> http://ldglite.sf.net/ldgliteosx1_2_3beta5.zip
Yep, that fixed it.
--Travis
| | | | | | |