Repository navigation
Conversation
The known-limitations note claimed the library falls back to SD card capture when streaming fails on the A2500. It does not: shoot(stream=True) goes to _shoot_streaming and raises, and nothing calls _shoot_standard after it. A caller reading that line would plan for a fallback that was never written.
|
Closing as superseded rather than rejected. The correction this PR makes — that nothing falls back from streaming capture, so The version that landed went further in a way that matters. This PR's text retains two claims that have since been checked and withdrawn: that the A2500 port is alpha-level, and that streaming fails with PTP Merging this now would reintroduce both. The branch |
The known-limitations note said the library falls back to SD card capture when streaming fails on the A2500. It does not.
shoot(stream=True)goes to_shoot_streamingand raises; nothing calls_shoot_standardafter it.A caller reading that line would plan for a fallback that was never written — which is how it surfaced: Captua's CHDK backend captures with
stream=Trueon the strength of it, and a review of that backend asked why it had no fallback path.The note now says what is true: streaming may fail with PTP
0x2002, it raises, and a caller that wants the picture anyway has to callshoot()itself and collect it from the card. It also says that whether the A2500 fails this way in practice has not been established against hardware — the cameras have not been run yet.