[Bug 736896] [REGRESSION] pygobject 3.13 now copies the GstStructure when getting them from a GstCaps, making it impossible to properly modify structures from caps in place.

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=736896

--- Comment #26 from Thibault Saunier <tsaunier at gnome.org> ---
(In reply to John Slade from comment #22)
> I have tested 3d19875eb7c20cedb4ab64e295c338ebd32f4b04 and it doesn't fix
> the problem of the sample being un-writable.

That was expected :)

> What do your patches in
> http://cgit.collabora.com/git/user/tsaunier/gst-python/ break?  I have been
> using them for a while and haven't found any problems.

It breaks rendering in Pitivi, I did not investigate further why tbh, from what
I can tell, PyGObject thinks it owns the boxed structure but it actually
doesn't with that patch.

(In reply to Nicolas Dufresne (stormer) from comment #24)
> My two cents. While this is being called a regression in the title,
> returning a copy of a boxed type with (transfer none) is a safe thing to do.
> Before that, it was really easy to crash the interpretor. Though, pygobject
> could rather put a "ref" to the original object to actually bind the
> lifetime (doing this at the Python reference level rather then touching the
> object references). Effectively creating a cirular reference though, which
> will hopefully be fixed by the GC.

Where and how do you imagine that to be done ?

(In reply to Will Manley from comment #25)
> > While this is being called a regression in the title
> 
> It is a regression because there were perfectly reasonable (and correct)
> programs that worked previously that no longer work.

I agree, this is a regression, it used to work but does not anymore.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.