#690 GUI gets frozen

[sistisette]

From time to time (I haven't been able to isolate a set of steps that reproduces this), an error message like the following appears on the console:

(Tcl) INVALID COMMAND NAME: invalid command name ".c"
while executing
".c coords a7e9d8RCOVER 869 255 877 333"
("uplevel" body line 1)
invoked from within
"uplevel #0 $cmd_from_pd"

and suddenly ALL the gui elements of a certain type or of several types (e.g. all sliders and/or all toggles and/or all number boxes) stop working, meaning that they will stop responding to mouse interaction and stop being updated.

Here's another example of the error message:

(Tcl) INVALID COMMAND NAME: invalid command name ".x1814f00.c"
while executing
".x1814f00.c itemconfigure .x17a9308.t1809158R -fill blue"
("uplevel" body line 9287)
invoked from within
"uplevel #0 $cmd_from_pd"

This happens to me from time to time with very complex patches with a lot of GOP-enabled abstractions.
I'm sorry I can't provide a simple test patch but i think this was worth reporting as it is a complete show stopper.

[Hans-Christoph Steiner]

An example patch is always very helpful, especially if it triggers the problem every time that it is run. Ideally the example patch would have only enough objects to cause the problem, but nothing else.

[Hans-Christoph Steiner]

Providing any test patch is better than nothing. The best test patch will reproduce the problem everytime. Its OK for it to be complicated if that is what it takes to trigger the problem.

Also, complete platform info is needed to narrow down the bug.

[sistisette]

Sorry for the missing platform information: it was Windows XP and Vista (tried both).

Unfortunately the (complicated) patch that (not always) reproduces the issue is not mine, so I cannot share it. I know this doesn't help very much but I thought the issue had to be reported and I hoped the complete error message would be a starting point.

[sistisette]

By the way, with 0.42.5 (on both Windows and Linux) I very very often get (and always got) error messages like these:
invalid command name ".xc11d790.c"
(without the additional information that I get on 0.43) but this usually doesn't cause any visible malfunctioning and i think it happens when closing patches.

I wonder whether it may be related

[sistisette]

Is this going to get serious attention? We are (and have been for a long time now) in a limbo where version 0.43 is unusable and version 0.42 does not get critical bugfixes or improvements because it is supposed to be an old version (but it is the latest stable one).
If this bug is not going to be fixed any soon, then version 0.43 should be put back into beta stage until at least critical blocker regressions like this are fixed.

[Hans-Christoph Steiner]

There are two errors here, the second one, GOP patches were calling the canvas name after the canvas has been deleted. That's fixed in both pure-data.git and pd-extended.git. The first one is caused by lines being wrongly cut in the middle. I think this is fixed in the latest nightly builds of Pd-extended and will be submitted to be included in pure-data.git after some testing.

[sistisette]

In which version is this supposed to be fixed?

I've tried today version 0.43-2 on Mac OS (the precompiled binary at download.puredata.info) and it still has this issue.

[Hans-Christoph Steiner]

This should be fixed in the latest Pd-extended 0.43.4. Is it still an issue for you?

[sistisette]

Not fixed at all in Extended.

I still experience this in Pd-Extended 0.43.4 on Linux. (i don't think this is platform depended as I observed the issue in all Linux, Windows and Mac since the beginning)

A long time ago M.Puckette said he believed this was fixed (but at least the first time he said so it wasn't) and then I later tried Vanilla 0.45 and I think I've never observed this any more in Vanilla >=0.45, though I may not have tested enough.

Isn't it worth checking whether it's fixed in Vanilla, and perhaps accelerate the process of getting Extended in sync with Vanilla? It's pretty irritating to always have Extended a couple of versions behind Vanilla.

Are these related?
https://sourceforge.net/p/pure-data/bugs/1021/
https://sourceforge.net/p/pure-data/bugs/1066/

The latter seems to include a reproducing patch (though I haven't tested that)

[Anonymous]

I can confirm that this is still a problem on Pd-extended 0.43.4 on OS X. I haven't isolated where in my complex patch it's happening, but I'm getting the "(Tcl) INVALID COMMAND NAME: invalid command name ".x....." type error and my GUI isn't working properly.

Phil Stone

[Anonymous]

Further information: the error seems to result from using the "playlist" object, which has not been maintained for years. Removing any occurrence of [playlist] from my patches has eliminated the error message and the gui malfunctioning.

Phil Stone

[Anonymous]

Each time I open "09.BinauralAmbisonicAndRoomsim.pd" from http://grh.mur.at/publications/sound-spatialization-pd
(http://grh.mur.at/sites/default/files/PdSpatialization.tar_.gz)
I get the error of the following type:

expr, expr~, fexpr~ version 0.4 under GNU General Public License

(Tcl) INVALID COMMAND NAME: invalid command name ".x360c00.c"
while executing
".x360c00.c itemconfigure 22cf400SRC1 -font {times 32 bold}"
("uplevel" body line 286)
invoked from within
"uplevel #0 $cmds_from_pd"matrix_inverse regular
AABCABCDEABCDEFGABCDEFGHI
0111222223333333444444444
sum of right channel: ++-++-+-++-+-+-++-+-+-+-+

All sliders get frozen.
I have OS X 10.9.5 and Pure Data 0.43.4-extended.