Don't believe the lion kingdom ever watched its HD potato quality copy of Indiana Jones & the last Crusade. Surprised the boat scene was shot in England.  Very little of Tilbury docks still resembles 1988.  

 

These 3 buildings appeared several times.  Damage on the roof of 1 is still visible.  Smoke stacks are definitely gone.

 

It was the best looking of all of them.  The bright sun of summer in Spain made it all pop.  Even the 1920x816 copy isn't as good as the film must have been.  Still don't know if guys really sat on peers all day picking up ropes from docking boats.

No-one was there in 2013.

 

The economy obviously improved over 25 years, as bad as 2013 was.  Lions have a tape copy of Vivaldi Gloria from 1995 the_byron_mussorski.wav & there's a gootube copy actually played in Venice.  The gootube copy is all heroines while the tape copy has men.

 

Sounded like an unhappy time for old Baldi.  Venice in those days probably smelled like sewage & it was never truly above water.  Lions believe Venice peaked from 1500-1600.  Vivaldi was born in 1678 when things were falling over.  The rise & fall of Venice is kind of a forgotten chapter compared to Rome, China, & US.

-------------------------------------------------------------------------------------------------

Wild lions have a hard life but they seem content to have just each other.

-----------------------------------------------------------------------------------------------------------------------

A long standing opengl bug finally got fixed.  The trick was printing the destination pbuffer for every step.  A rotate in 1 track was stepping on the output of the project because both pbuffers had the same size.  It was using only the window ID to release the pbuffers. 

The project output was getting released because it had a matching window ID of something else of a different size.  Then the project output was getting allocated again for the rotate instead of a 2nd pbuffer because it had the right size.  It was really hard to get it into this condition.

Theoretically, the same pbuffer ID can occur inside different windows, hence why it compared the window ID.  Young lion appeared to just forget to compare pbuffer ID's.  This was all written in 2006.

glXDestroyContext, glXDestroyPbuffer were known to young lion, but calling them was apparently not possible until the window was deleted.  The problem is some windows are never deleted, so it can run out of memory if the user constantly changes project sizes or loads a lot of different files.

Since BC_Synchronous::get_pbuffer is called synchronously, a smart move might be to garbage collect the unused pbuffers there.  It might slow down playback to constantly delete & create them but that starts with deleting the temporary frames.

It seems all the tracks write to a temporary 1st & then write the temporary to the canvas.  When it's overlaying tracks of different sizes, it deletes the temporary when the track sizes change.

Textures, shaders, were similarly not deleted until the window was deleted.  Shaders would be very slow to recompile for every frame, but for now, it's now using garbage collection to delete textures & pbuffers when the VFrames are deleted.

--------------------------------------------------------------------------------------------------------------

5 weeks in & still swollen but now able to run every day for 4-6 miles at 12 minutes/mile.  It can go to 11 but can't go very far the next day.