[+] add details what should be configured and when
[ ] how using the custom_clock patch allows to take some snapshots, but teorically weakens timeline comparisons; to be strict a snapshot key should report if a patched custom_clock was used
[+] warn about setting the pillow dependency on conf.py
[+] warn about letting the console have focus
[ ] mention limitations on going too much back in time re compat test\ vs cocos lib under test
[ ] mention it is uncertain which pyglet version was tested for certains old cocos releases: long hiatus in pyglet releases, confusing pyglet history at the time by merging multiple long lived branches. Date is a hint.
[ ] mention small expected diferences. here a copy of annotations of test runs
pyglet 1.5.8 , cocos 0.6.9 pre release (master 6be712d "revert test_pyglet_vb.py so compatible with older cocos - testing")
v_ref = ("3.7", "a39ebc20", "master")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8")
-> diferencias usuales mas una diferencia un poco mas notable en test_pyglet_vb
pyglet 1.5.8 , cocos 0.6.9 pre release (master 6be712d "revert test_pyglet_vb.py so compatible with older cocos - testing")
v_ref = ("3.8", "a39ebc20", "master")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8")
-> reporta un traceback y es
Details snapshots_failure errors
test_pyglet_vb.py
Error in sys.excepthook:
Original exception was:
El traceback aparecio con py 38; en 3.7 no estaba, creo que son cambios en exceptions dentro de exceptions
Despues solo hay las diferencia usuales
@1
pyglet 1.5.2 , cocos 0.6.8
v_ref = ("3.8", "edf43ad5", "release-0.6.8")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8")
-> 2 tracebacks, en los sin tracebacks no diferencias (excepto la usual en test_action_non_interval)
Aunque reporta 2 tracebaks, en la sección 'Details about tests that failed to take all snapshots', que es fonde veiamos los tracebacks, solo lista un script, test_pyglet_vb.py
pyglet 1.5.2 , cocos 0.6.8
v_ref = ("3.7", "edf43ad5", "release-0.6.8")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8")
-> no tracecack, delta: no diferencias (excepto la usual en test_action_non_interval), ok
pyglet 1.4.11 (last pyglet 1.4.x), cocos 0.6.7 (1st cocos version to support pyglet 1.4.x)
v_ref = ("3.7", "0af100f4", "release-0.6.7")
pyglet 1.4.11 (last pyglet 1.4.x),
v_other = ("3.7", "0af100f4", "release-0.6.8")
-> no tracecack, delta: no diferencias, ok,
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.6 (last cocos version to support pyglet 1.3.x)
v_ref = ("3.7", "0419b96c", "release-0.6.6")
pyglet 1.4.11 (last pyglet 1.4.x),
v_other = ("3.7", "0af100f4", "release-0.6.8")
-> no tracecack, delta: small, ok, test_all_collisions qyuzas invertido el orden de dibujado de los circulos
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5 (1st cocos version to support pyglet 1.3.x)
v_ref = ("3.6", "0419b96c", "release-0.6.5")
pyglet 1.4.11 (last pyglet 1.4.x),
v_other = ("3.7", "0af100f4", "release-0.6.8")
-> no tracecack, delta: small, ok
pyglet last 1.2-maintenance, cocos 0.6.4
v_ref = ("3.6", "a9c6ffdb", "release-0.6.4")
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5 (1st cocos version to support pyglet 1.3.x)
v_other = ("3.6", "0419b96c", "release-0.6.5")
-> no traceback, dos diferencias notables:
todos los textos parecen tener pequeñas diferencias en los bordes i/o posicion
en algunos menues falta el titulo -> veifique manualmente es asi
test_batch3.py : parece que dos planos de sprites son dibujados en diferentes orden
test_action_non_interval.py : peqhenos deltas en los bordes de quads, probablemente debidos a pequeña dpos
despues hay los pequeños deltas habituales de bordes de alpha.
pyglet last 1.2.x released, 1.2.4, cocos 0.6.4
v_ref = ("3.6", "2f2283d6", "release-0.6.4")
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5 (1st cocos version to support pyglet 1.3.x)
v_other = ("3.6", "0419b96c", "release-0.6.5")
-> los titulos de menues estan bien, que en el anterior estaban mal
eb test_batch3.py hay diferencias notables: aparentememte una capa de sprites no se renderizo
------------------ ruptura de compat entre los tests y la lib. Arriba podemos usar los tests de cocos 0.6.9, abajo
------------------ hay que usar otro set, posiblemente de 0.6.3
pyglet last 1.2.x released, 1.2.4, cocos 0.6.3
v_ref = ("3.6", "2f2283d6", "release-0.6.3")
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5
v_other = ("3.6", "0419b96c", "release-0.6.5")
-> tracebacks en release-0.6.3
ModuleNotFoundError: No module named 'cocos.scenes.sequences' -> parte de algun refactor
test_jumptiles3d.py -> cocos\actions\tiledgrid_actions.py", line 436, in update
coords[k+2] += sinz2
NameError: name 'sinz2' is not defined -> a bug from some reformat
test_platformer.py
from cocos import tiles, actions, layer, mapcolliders
ImportError: cannot import name 'mapcolliders' -> de un refactor
test_scrolling_manager_without_tiles.py
test_scrolling_manager_without_tiles_autoscale.py
screen_pos = self.scroller.world_to_screen(*position)
AttributeError: 'ScrollingManager' object has no attribute 'world_to_screen'-> de un refactor
-> Diferencias gordas:
test_tiles.py: tiles misma posicion, autito desplazado
test_tiles_autotest: hay un deplazamiento horizontal entre ambas imagenes
test_batch3: cambio de render order entre dos batches
Cambios chicos
En todos los de particulas hay algo de diferencias, poco perceptibles en general, quizas se nota mas en test_particle_explosion que pareceria mostrar una pequeña diferencia en tamaño de particulas.
[+] add small overview intention and mechanics
[+] add details what should be configured and when
[ ] how using the custom_clock patch allows to take some snapshots, but teorically weakens timeline comparisons; to be strict a snapshot key should report if a patched custom_clock was used
[+] warn about setting the pillow dependency on conf.py
[+] warn about letting the console have focus
[ ] mention limitations on going too much back in time re compat test\ vs cocos lib under test
[ ] mention it is uncertain which pyglet version was tested for certains old cocos releases: long hiatus in pyglet releases, confusing pyglet history at the time by merging multiple long lived branches. Date is a hint.
[ ] mention small expected diferences. here a copy of annotations of test runs
===================== maq win 10 ======================================================
pyglet 1.5 dev 2020 10 28, cocos 0.6.9 pre release (master 6be712d "revert test_pyglet_vb.py so compatible with older cocos - testing")
v_ref = ("3.7", "b1b2e5e1", "master")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8") -> las diferencias menores de siempre; acá mostró mas diferencias en los fps
pyglet 1.5 dev 2020 10 28, cocos 0.6.9 pre release (master 8367d9635 "more fix tiles for py 3.9+ xml changes")
v_ref = ("3.9", "b1b2e5e1", "8367d9635")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8") -> ok, solo las pequeñas diferencias, test_pyglet_vb un poco mas de diferencia.
====================== abajo es maq win 7 ======================================================
pyglet 1.5 dev 2020 10 28, cocos 0.6.9 pre release (master 6be712d "revert test_pyglet_vb.py so compatible with older cocos - testing")
v_ref = ("3.7", "b1b2e5e1", "master")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8") -> diferencias usuales
pyglet 1.5.8 , cocos 0.6.9 pre release (master 6be712d "revert test_pyglet_vb.py so compatible with older cocos - testing")
v_ref = ("3.7", "a39ebc20", "master")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8") -> diferencias usuales mas una diferencia un poco mas notable en test_pyglet_vb
pyglet 1.5.8 , cocos 0.6.9 pre release (master 6be712d "revert test_pyglet_vb.py so compatible with older cocos - testing")
v_ref = ("3.8", "a39ebc20", "master")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8") -> reporta un traceback y es Details snapshots_failure errors test_pyglet_vb.py Error in sys.excepthook: Original exception was: El traceback aparecio con py 38; en 3.7 no estaba, creo que son cambios en exceptions dentro de exceptions
Despues solo hay las diferencia usuales
@1
pyglet 1.5.2 , cocos 0.6.8
v_ref = ("3.8", "edf43ad5", "release-0.6.8")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8") -> 2 tracebacks, en los sin tracebacks no diferencias (excepto la usual en test_action_non_interval) Aunque reporta 2 tracebaks, en la sección 'Details about tests that failed to take all snapshots', que es fonde veiamos los tracebacks, solo lista un script, test_pyglet_vb.py
pyglet 1.5.2 , cocos 0.6.8
v_ref = ("3.7", "edf43ad5", "release-0.6.8")
pyglet 1.4.11 (last released 1.4.x), cocos 0.6.8
v_other = ("3.7", "0af100f4", "release-0.6.8") -> no tracecack, delta: no diferencias (excepto la usual en test_action_non_interval), ok
pyglet 1.4.11 (last pyglet 1.4.x), cocos 0.6.7 (1st cocos version to support pyglet 1.4.x)
v_ref = ("3.7", "0af100f4", "release-0.6.7")
pyglet 1.4.11 (last pyglet 1.4.x),
v_other = ("3.7", "0af100f4", "release-0.6.8") -> no tracecack, delta: no diferencias, ok,
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.6 (last cocos version to support pyglet 1.3.x)
v_ref = ("3.7", "0419b96c", "release-0.6.6")
pyglet 1.4.11 (last pyglet 1.4.x),
v_other = ("3.7", "0af100f4", "release-0.6.8") -> no tracecack, delta: small, ok, test_all_collisions qyuzas invertido el orden de dibujado de los circulos
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5 (1st cocos version to support pyglet 1.3.x)
v_ref = ("3.6", "0419b96c", "release-0.6.5")
pyglet 1.4.11 (last pyglet 1.4.x),
v_other = ("3.7", "0af100f4", "release-0.6.8") -> no tracecack, delta: small, ok
pyglet last 1.2-maintenance, cocos 0.6.4
v_ref = ("3.6", "a9c6ffdb", "release-0.6.4")
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5 (1st cocos version to support pyglet 1.3.x)
v_other = ("3.6", "0419b96c", "release-0.6.5") -> no traceback, dos diferencias notables: todos los textos parecen tener pequeñas diferencias en los bordes i/o posicion en algunos menues falta el titulo -> veifique manualmente es asi test_batch3.py : parece que dos planos de sprites son dibujados en diferentes orden test_action_non_interval.py : peqhenos deltas en los bordes de quads, probablemente debidos a pequeña dpos despues hay los pequeños deltas habituales de bordes de alpha.
pyglet last 1.2.x released, 1.2.4, cocos 0.6.4
v_ref = ("3.6", "2f2283d6", "release-0.6.4")
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5 (1st cocos version to support pyglet 1.3.x)
v_other = ("3.6", "0419b96c", "release-0.6.5") -> los titulos de menues estan bien, que en el anterior estaban mal eb test_batch3.py hay diferencias notables: aparentememte una capa de sprites no se renderizo
------------------ ruptura de compat entre los tests y la lib. Arriba podemos usar los tests de cocos 0.6.9, abajo ------------------ hay que usar otro set, posiblemente de 0.6.3
pyglet last 1.2.x released, 1.2.4, cocos 0.6.3
v_ref = ("3.6", "2f2283d6", "release-0.6.3")
pyglet 1.3.3 (last released 1.3.x), cocos 0.6.5
v_other = ("3.6", "0419b96c", "release-0.6.5") -> tracebacks en release-0.6.3 ModuleNotFoundError: No module named 'cocos.scenes.sequences' -> parte de algun refactor
-> Diferencias gordas:
test_batch3: cambio de render order entre dos batches
Cambios chicos
En todos los de particulas hay algo de diferencias, poco perceptibles en general, quizas se nota mas en test_particle_explosion que pareceria mostrar una pequeña diferencia en tamaño de particulas.
Diferencias menores: