Re: [gob-list] export wrappers for GObject and Boxed types

From: George <>
Date: Tue, 26 Dec 2006 18:26:21 -0800

On Sun, Dec 24, 2006 at 06:51:50AM -0600, Michael Lawrence wrote:
> Dear GOB people,
> Is anyone else annoyed by the autogenerated "export" wrappers for properties
> using g_object_get(), which results in increasing the reference count /
> copying of objects and boxed types? While it is common to copy strings, for
> instance, copying boxed types and reffing objects is somewhat unconventional
> relative to GTK+ and other GObject-based APIs. Could there be an extra flag
> that specifies whether the caller should be given ownership of an exported
> property value? In my opinion, that would make the "export" keyword a lot
> more useful.

As far as I remember, the consensus was that the verb 'get' meant reffing (or
copying). If an option would be added to make it possible to get at the
property value without reffing and copying, it should generate a 'peek'
function rather then a 'get' function.

> By the way, has GOB development stalled? Another feature I am looking
> forward to is the definition of interface types.

Well, I'm no longer actively coding anything. I'm maintaining gob as far as
it needs but since I'm not using it currently (I'm not really coding
anything) I likely won't find any errors or need for new features. It seems
like enough people are using gob, but most people seem satisfied enough with
the featureset it has (enough not to code the features themselves).

There was someone some time ago who wanted to rewrite gob completely but
nothing seems to have happened on that.

IMO adding the peek option as you want above should be a few hours work at
most and definition of interface types is probably a weekend or two. I
likely won't do either anytime soon, but of course this is free software,
and I'm accepting patches!


George <>
   History teaches us that men and nations behave wisely
   once they have exhausted all other alternatives.
                       -- Abba Eban, 1970
Received on Wed Dec 27 2006 - 02:26:49 CST

This archive was generated by hypermail 2.2.0 : Sun Apr 17 2011 - 21:05:02 CDT