[Bug 667352] [0.11] misc core API/bug comments/nitpicks
bugzilla at gnome.org
Tue Mar 27 06:27:42 PDT 2012
GStreamer | gstreamer (core) | 0.11.x
--- Comment #8 from Wim Taymans <wim.taymans at gmail.com> 2012-03-27 13:27:37 UTC ---
(In reply to comment #6)
> * gst_buffer_span has disabled code that will copy over _OFFSET etc if the
> offset == 0. Not doing so impacts gst_buffer_join in particular and affects
> all sorts of elements in subtle ways (e.g. makes the icydemux test fail etc).
> Don't know what's the deal/idea here; re-enable ? or is there an alternative.
I fixed it now, no idea why it was disabled.
> * seems like elements should be very careful these days to return FALSE when
> receiving some event (e.g. a newsegment event in an unexpected format), since
> upstream will keep trying to send it (if sticky, as many are) and fail hard
> when "real data" is then being pushed (and the sticky event still was not
> received properly). Maybe self-evident, but probably a problem here and there
> for elements using an event return value for some hacks that might then need (a
> query?) alternative (e.g. flacenc springs to mind).
Refusing an event will now also generate an error when trying to push a buffer.
I think this makes sense in the case of a segment event with a wrong format.
Will check what hack flacenc does.
> * crop metadata vs caps:
> it looks like the caps width/height should reflect the image's total
> width/height, and not the cropped width/height (presumably so an element not
> having such crop meta support can carry on as usual). If indeed so, maybe
> having this specified/documented somewhere?
The caps should reflect what the user visible region is, so _after_ cropping.
The idea is that a capsfilter still affects the visual result. An element can
only send a crop metadata if downstream understands, that means all the
downstream elements should understand (and explicitly enable) the crop
metadata. If downstream does not accept the crop metadata, the upstream element
software cropping (which might not be very accurate).
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.
More information about the gstreamer-bugs