[bug #45833] support load/save of classdef objects

Please use the bug tracker to post updates to a bug report.  The mailing list is intended as a read-only notification stream.  Info posted to this mailing list address won't appear in the tracker database where it is most useful.

Follow-up Comment #96, bug #45833 (group octave):

It's a great question, I haven't really thought about it before now. So all
MCOS metadata (within Filewrapper or otherwise) are saved as uint32 vectors,
which is why I defaulted to uint32. However I very recently learnt that class
and property names are indexed by signed ints.

This would mean that some specific type of metadata requires the whole array
to be unsigned ints. It is hard to reverse engineer this behaviour, but I
speculate its either because of (i) the MCOS identifier (0xDD000000) or (ii)
property and class names in the metadata (although I believe this should be
strictly ASCII and hence unlikely).

I think we can safely use signed ints for most of the variables in the
warnings below. The only place where I doubt is `m_region_offsets". This is
supposed to contain integer offsets to different blocks in the same metadata
vector, which I think would be better off using an unsigned int. But again
this behaviour is hard to verify against MATLAB.


    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?45833>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/

-----BEGIN PGP SIGNATURE-----

iHUEABYIAB0WIQQk97aszIMMAvLLwm6qLAuaBUf3TgUCaYYqJwAKCRCqLAuaBUf3
Tnw7AQCZU/2q4wJ6IWd+Tpxff2pgh0bRcjcbJBvQ87NdlPSJdwD+P1L97pum2W9V
R4sjHoUWtjGX8IykofZI9l1aph0Huww=
=7MK5
-----END PGP SIGNATURE-----