Pranav Kant <> changed:

           What    |Removed                     |Added
             Status|UNCONFIRMED                 |ASSIGNED
           Assignee| |
                   |                 |
     Ever confirmed|0                           |1

--- Comment #1 from Pranav Kant <> ---
(In reply to Mike Gorse from comment #0)
> User-Agent:       Mozilla/5.0 (X11; Linux x86_64; rv:48.0) Gecko/20100101
> Firefox/48.0
> Build Identifier: 10m0(Build:2)
> Libreoffice will accept versions of gobject-introspection as far back as
> 1.32.0 when configuring. However, LibreOfficeKit uses the (nullable)
> annotation, which was added in the 1.42 cycle. So, if an older version is
> installed, then introspection data for these functions will be generated
> without (nullable) for these functions, without any obvious indication that
> something has gone wrong (aside from possibly generating a warning when
> g-ir-scanner is run). This leads to unexpected behavior. For instance,
> lok_doc_view_new() might be called with an empty pPath string, rather than a
> null string, and the function does not check for an empty string and treat
> it as equivalent to NULL. We could do one of the following:
> (1) replace (nullable) with (allow-none); the latter is supported by older
> versions of gobject-introspection and is equivalent to (nullable) for in
> parameters.

allow-none is deprecated since 1.42; How about putting both allow-none and
nullable together wherever required to keep the backward compatibility upto
1.32 and then we can remove (allow-none) when we bump the version in future.

My quick tweaking session with .gir and .typelib files seems to suggest that
this should fix the issue, so hopefully this should be sufficient to address
the crash. However, if we face more problems other than this, I think better to
just bump the GI version.

You are receiving this mail because:
You are the assignee for the bug.
Libreoffice-bugs mailing list

Reply via email to