To tell you the truth, I don't understand the IEM code very well, and I worry about the prospect of eventually having to port it to some other GUI toolkit besides TK (assuming that indeed turns out to be the main graphics performance problem in Pd, which I'm less and less sure of these days!)

I'm certainly scared of having to drag yet more code around with pd; one of my main aims is long-term code stability.

cheers Miller

On Sun, Jan 18, 2004 at 10:47:07AM +0100, Frank Barknecht wrote:

Hallo, Miller Puckette hat gesagt: // Miller Puckette wrote:

...

I think I've got them straight now...

I have the same problem with the knob external which I hacked together some time ago as an almost verbatim copy of th v/hsl code in the IEMGUIs. This means: I really don't understand the code. :(

Could you give some pointers on what is changing in the GUIs currently, so I might adapt knob, too? And maybe: Would you consider putting something like knob into Pd itself?

ciao Frank

...

On Tue, Jan 13, 2004 at 05:09:37PM -0800, ben@ekran.org wrote:

...

Hello Miller,

I found a bug to do with canvas names in pd 0.37-1 test2 and test3 (I'm compiling 37-0 now)

The send/receive and name params seem to be all mixed up.

If I create a canvas with the send "1" receive "2" and name "3" close and open the patch I get a canvas with send "1" receive "2" and name "2" (the receive param seems to be copied into the name param.)

Ok so there is this issue so I'm trying to compile a working pd but...

37-0 fails with:

d_ctl.c: In function `line_tilde_perform': d_ctl.c:110: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c:110: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c: In function `vline_tilde_float': d_ctl.c:285: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c:285: warning: dereferencing type-punned pointer will break strict-aliasing rules make: *** [d_ctl.o] Error 1

36-0 fails with:

g_bang.c: In function `bng_new': g_bang.c:442: warning: dereferencing type-punned pointer will break strict-aliasing rules g_bang.c:443: warning: dereferencing type-punned pointer will break strict-aliasing rules make: *** [g_bang.o] Error 1

using gcc3.3

Sounds a little too familar! Any ideas for a fix before I wreak all my canvases!!!?

Thanks Ben

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev

-- Frank Barknecht _ ______footils.org__

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev

On Sun, 18 Jan 2004, Miller Puckette wrote:

besides TK (assuming that indeed turns out to be the main graphics performance problem in Pd, which I'm less and less sure of these days!)

I think this is really hard to tell. Assuming the implementation of the canvas widget is more or less optimal (except on OSX probably), one additional overhead that is surely added is the parsing and interpreting of the tcl messages that are sent to the gui. But to which amount is this responsible for the GUI problems, if at all ?

Guenter

I'm certainly scared of having to drag yet more code around with pd; one of my main aims is long-term code stability.

cheers Miller

On Sun, Jan 18, 2004 at 10:47:07AM +0100, Frank Barknecht wrote:

...

Hallo, Miller Puckette hat gesagt: // Miller Puckette wrote:

...

I think I've got them straight now...

I have the same problem with the knob external which I hacked together some time ago as an almost verbatim copy of th v/hsl code in the IEMGUIs. This means: I really don't understand the code. :(

Could you give some pointers on what is changing in the GUIs currently, so I might adapt knob, too? And maybe: Would you consider putting something like knob into Pd itself?

ciao Frank

...

On Tue, Jan 13, 2004 at 05:09:37PM -0800, ben@ekran.org wrote:

...

Hello Miller,

I found a bug to do with canvas names in pd 0.37-1 test2 and test3 (I'm compiling 37-0 now)

The send/receive and name params seem to be all mixed up.

If I create a canvas with the send "1" receive "2" and name "3" close and open the patch I get a canvas with send "1" receive "2" and name "2" (the receive param seems to be copied into the name param.)

Ok so there is this issue so I'm trying to compile a working pd but...

37-0 fails with:

d_ctl.c: In function `line_tilde_perform': d_ctl.c:110: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c:110: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c: In function `vline_tilde_float': d_ctl.c:285: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c:285: warning: dereferencing type-punned pointer will break strict-aliasing rules make: *** [d_ctl.o] Error 1

36-0 fails with:

g_bang.c: In function `bng_new': g_bang.c:442: warning: dereferencing type-punned pointer will break strict-aliasing rules g_bang.c:443: warning: dereferencing type-punned pointer will break strict-aliasing rules make: *** [g_bang.o] Error 1

using gcc3.3

Sounds a little too familar! Any ideas for a fix before I wreak all my canvases!!!?

Thanks Ben

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev

-- Frank Barknecht _ ______footils.org__

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev

I think I've fixed one problem with canvas naming... but I don't have GCC 3.3 running anywhere here so can't easily track down those problems. (although the ones below are fixed, I think, in 0.37-1test3.)

The makefile no longer has "-Werror" by default, so you can at least get Pd compiled without editing the makefiles. Those warnings are real bugs, though, and if you get them stuff might not work right!

cheers Miller

On Sun, Jan 18, 2004 at 07:26:58AM -0800, ben@ekran.org wrote:

Great!

Does this include both the gcc3.3 failures and the Canvas name issues?

As for compiling with gcc3.3 I have removed the -Werror flags from the makefiles, but the compile still fails...

Thanks Ben

...

...

Ok so there is this issue so I'm trying to compile a working pd but...

37-0 fails with:

d_ctl.c: In function `line_tilde_perform': d_ctl.c:110: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c:110: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c: In function `vline_tilde_float': d_ctl.c:285: warning: dereferencing type-punned pointer will break strict-aliasing rules d_ctl.c:285: warning: dereferencing type-punned pointer will break strict-aliasing rules make: *** [d_ctl.o] Error 1

36-0 fails with:

g_bang.c: In function `bng_new': g_bang.c:442: warning: dereferencing type-punned pointer will break strict-aliasing rules g_bang.c:443: warning: dereferencing type-punned pointer will break strict-aliasing rules make: *** [g_bang.o] Error 1

using gcc3.3

Sounds a little too familar! Any ideas for a fix before I wreak all my canvases!!!?

Thanks Ben

---

PD-dev mailing list PD-dev@iem.at http://iem.at/cgi-bin/mailman/listinfo/pd-dev