Re: the nvidia/osmesa dilema

[Daniel J Sebald]

On 05/04/2016 10:56 AM, John W. Eaton wrote:
> On 05/04/2016 11:49 AM, Mike Miller wrote:
> > On Wed, May 04, 2016 at 10:43:34 -0500, Daniel J Sebald wrote:
>
> > > That, or fix the issue with nVidia off-screen printing.
> >
> > True.
>
> Yes, obviously this would be the best solution.  Can you help?  Please
> take a look at the most recent 4 comments posted to the following bug
> report, for example:
>
>    https://savannah.gnu.org/bugs/?47372

Is the above the same issue as that in 
https://savannah.gnu.org/bugs/?44478 ?  The former reads like a more 
particular bug, the latter more an issue with initializing the 
OpenGL-FLTK environment on a multi-threaded system.  At least that is 
what I understand.

Could someone summarize what the issue is?

If the core problem with building documentation is that there is no 
graphics initialization in the thread doing the graphics commands 
because the GUI isn't launched during build (I think that is what I'm 
getting from the posts at ?44478), then perhaps the solution is to 
figure out a way to use the graphics code in a single thread if the GUI 
is not launched.  And maybe that makes sense, don't know.  I mean, if 
someone is doing a non-native build on computer A (non-threaded) for 
computer B (threaded) then the graphics images couldn't be built for the 
documentation on computer A.  Of course, the odds of someone even owning 
computer A these days are next to nil--but it is the idea of Octave not 
requiring multiple threads if not in GUI mode, if that is something to 
aim for.

Dan