Hi, The global is now set in the lowest event code, to make it work for non-threading processes. I suggest to keep that, but make sure it only gets checked in case you use a blocking process.
Next to that, each running modal handler and WM job should get their own ESC handler, which can be coded in a way that it will escape the last added first. We can also add a notification for ended jobs in the header. It's a bit of coding work though... I could check on doing this for at least the common cases I know well (file window ESC, renders, anim play). -Ton- -------------------------------------------------------- Ton Roosendaal - [email protected] - www.blender.org Chairman Blender Foundation - Producer Blender Institute Entrepotdok 57A - 1018AD Amsterdam - The Netherlands On 24 May, 2013, at 2:33, Brecht Van Lommel wrote: > A more general solution is indeed needed here. The ESC key should go > through the typical event handling loop before being used to cancel a > running job, so that SCREEN_OT_animation_cancel, the operator to hide > the file browser, and similar operators can catch it before anything > else does. > > I think this could be implemented relatively easily with a "cancel > job" operator at the window level bound to the ESC key. This can then > set the G.is_break global if any jobs are running. That global should > really be removed and the cancel signal passed to a specific job, but > maybe that's more work than you want to invest in this. > > When moving this to an operator, a problem is that blender_test_break > needs to catch the ESC event when the event handling loop is skipped > due to some heavy job blocking the main thread. Actually > blender_test_break should get removed, it's used now just for point > cache and dynamic paint baking. The proper solution is to make those > non-blocking using the jobs system, but if I remember correctly that's > problematic because editing/drawing the scene will mess up the bake, > so that may not be so simple. > > The quick solution would be to add a separate global variable for > blender_test_break, and clearly mark blender_test_break as being > intended for operations that block the main thread. > > On Fri, May 24, 2013 at 1:07 AM, PabloVazquez.org <[email protected]> wrote: >> As a user, I've ran into this many times before as well. >> >> But not only on Alt+A, also while file browsing for example. Pressing Esc >> to close the file browser would stop my render. Is pretty common that while >> doing long renders you keep working on the .blend file doing other stuff. >> >> There might be other cases, maybe a more generic way to deal with this is >> needed? >> >> >> -- >> Pablo Vazquez >> CG Artist - Blender Foundation Certified Trainer >> Twitter: @venomgfx <http://twitter.com/venomgfx> >> Blender Network: blendernetwork.org/pablovazquez >> Email: [email protected] >> Website: pablovazquez.org <http://www.pablovazquez.org> >> >> >> On Thu, May 23, 2013 at 10:21 PM, Daniel Genrich >> <[email protected]>wrote: >> >>> Hey, >>> >>> I've come upon a very weird problem when using the ESC key. >>> >>> When I am e.g. baking a fluid simulation, then press alt-a (3dview >>> animation playback loop) and then press "ESC" the animation playback in >>> the 3dview is stopped BUT also is the fluidsim bake. This happens with >>> alot of stuff too, like renderer, compositor, etc. >>> >>> They all have in common that they rely on the "G.is_break" variable >>> which is set when ESC is pressed (wm_event_system.c line 3111). >>> In interface_templates.c (line 2817) we have a nice differentiated >>> function which could serve as example. >>> >>> In the moment it is very easy to loose valueable work by accidently >>> pressing ESC because you want to stop the 3dview animation playback . >>> >>> My proposal for wm_event_system.c line 3111: >>> >>> if(ESC_PRESSED) >>> { >>> if(animation_is_running) >>> { >>> WM_jobs_stop(CTX_wm_manager(C), CTX_data_scene(C), NULL); // or >>> event B_STOPANIM in do_jobs_running() >>> } >>> else >>> { >>> G.is_break = TRUE; >>> } >>> >>> >>> >>> Looking forward to your feedback, >>> Daniel >>> _______________________________________________ >>> Bf-committers mailing list >>> [email protected] >>> http://lists.blender.org/mailman/listinfo/bf-committers >>> >> _______________________________________________ >> Bf-committers mailing list >> [email protected] >> http://lists.blender.org/mailman/listinfo/bf-committers > _______________________________________________ > Bf-committers mailing list > [email protected] > http://lists.blender.org/mailman/listinfo/bf-committers _______________________________________________ Bf-committers mailing list [email protected] http://lists.blender.org/mailman/listinfo/bf-committers
