jreality doesn't start reliably

Questions and problems about using polymake go here.
beew
Posts: 30
Joined: 13 Aug 2015, 20:41

Re: jreality doesn't start reliably

Postby beew » 22 Nov 2015, 01:28

Hi, Thanks for the patch. I rebuilt polymake-beta with the patch but still doesn't work. Here is the new debug log

Also starting polymake normally I get this warning "Loading applications now...polymake: WARNING: undocumented category /objects/Cone/specializations/Cone<Rational>/properties/Algebraic Geometry at /usr/share/polymake/apps/fulton/rules/toric_ideal.rules, line 59"

I am not sure if it is because of the patch or something else. But it doesn't show up if start polymake in debug mode.
Attachments
debug.log
(77.82 KiB) Downloaded 289 times

beew
Posts: 30
Joined: 13 Aug 2015, 20:41

Re: jreality doesn't start reliably

Postby beew » 01 Dec 2015, 01:21

Hello? Any update?

blorenz
Developer
Posts: 139
Joined: 10 Jan 2011, 17:21

Re: jreality doesn't start reliably

Postby blorenz » 02 Dec 2015, 11:15

Hi,

we will continue investigating this problem but we currently have no further ideas why this happens. Thanks for your help and logs.

The only thing I can suggest right now is to use a different visualization backend, e.g.:

Code: Select all

threejs(cube(3)->VISUAL);
You can also set the default via: prefer "threejs";

Best,
Benjamin

PS: The warning is just related to a minor documentation issue and in debug mode it also appears but is redirected to the debug.log file.

beew
Posts: 30
Joined: 13 Aug 2015, 20:41

Re: jreality doesn't start reliably

Postby beew » 03 Dec 2015, 12:11

Hi,

I have build polymake 2.14r1 this time with the patch (instead of the beta), it sort of works better than before, jreality does start half of the time (before it never started) and the error message is also gone. But when jreality doesn't start killing java and rerun the command sometimes work, and if not deleting the config folder .polymake and restart polymake usually works. Not sure if this info is helpful.

Also as a suggestion, right now if I run a command like "$q -> VISIUAL;" polymake will jump to the next command prompt whether jreality starts or not. Sometimes it takes a while to start, sometimes never. There is no clue from looking at the terminal whether the command is successful. One can wait a while, jreality may open, or may not. I think it would be good if there is some clue, like the next prompt doesn't appear until jreality is successfully launched or output an error if it get stuck somehow.

Thanks.

P.S. I am using geomview for the viewer, works very well.

User avatar
gawrilow
Main Author
Posts: 423
Joined: 25 Dec 2010, 17:40

Re: jreality doesn't start reliably

Postby gawrilow » 03 Dec 2015, 12:52

I think it would be good if there is some clue, like the next prompt doesn't appear until jreality is successfully launched or output an error if it get stuck somehow.
The next prompt appears after java has successfully started as a background process, launched the jreality application, and it has established the connection over an internal socket. There is nothing more that polymake could sense distantly. The reason for jreality windows not appearing is obviously some bug or race condition within jreality itself or jogl libraries; the socket is solely used for raw data exchange like coordinate matrices or color attributes but no stuff related to GUI.

beew
Posts: 30
Joined: 13 Aug 2015, 20:41

Re: jreality doesn't start reliably

Postby beew » 08 Dec 2015, 14:01

Hi,

I found the solution! It turns out to be this java bug on Ubuntu/Debian https://bugs.launchpad.net/ubuntu/+sour ... ug/1510009
It works now after commenting out the single line in /etc/java-8-openjdk/accessibility.properties. I am using polymake 2.14r1 with blorenz's patch (maybe it would work without the patch too, but haven't tested it)


Return to “Helpdesk”