source/blender/makesrna/intern/rna_sequencer.c
166	How's `ID_RECALC_SEQUENCER_STRIPS` relevant here? This comment states something which is not clear from the context, it should give some more context information to be useful. Tagging is cheap, so you can as well do `ID_RECALC_AUDIO \| ID_RECALC_SEQUENCER_STRIPS` to avoid relying on a specifics of how `rna_Sequence_invalidate_raw_update` works.
167	This tag is to be done if old scene also was using audio. Simpler and more reliable is to just to always tag for `ID_RECALC_AUDIO \| ID_RECALC_SEQUENCER_STRIPS`.
170–171	Relations update doesn't necessarily evaluate anything, making the comment irrelevant/wrong. The proper reasoning here is that the scene which used to be used by the strip is to be removed from the depsgraph and the new one is to be added. But stating such things in all usages of `DEG_relations_tag_update()`will be more of unnecessary over-documentation.
2244	The convention is to use `_update` suffix for update callbacks. Would be something like: `rna_Sequence_strip_scene_update`

This revision now requires changes to proceed.Oct 21 2019, 10:32 AM

source/blender/makesrna/intern/rna_sequencer.c
166	My reasoning was, that sure, tagging is cheap, but it can trigger recalc, which is not. So I wanted to tag `ID_RECALC_AUDIO` only when I expect audio to be actually used(reproduced) Perhaps it's better to tag `ID_RECALC_AUDIO \| ID_RECALC_SEQUENCER_STRIPS` in this case, as this func is run only when switching scenes.
170–171	I left this comment here to validate my assumption. So if I understand you correctly, does it mean, that we should call `DEG_relations_tag_update()` each time any ID datablock usage has been changed?