Re: initializing extensions in daily from Master slooow

On Wed, 2012-03-28 at 21:25 -0500, Norbert Thiebaud wrote:
> or you could actually call sync() directly from the save() method once
> your done with your writing (*),

        Urgh; wow, synching the whole system (though fsync tends to end up
doing this on ext2/ext3 anyway) is also a bit extreme, certainly
inelegant and osl (luckily) has no abstraction for that ;-)

> that push up some implementation details... but that still would be a
> less ugly than the hack above

        Well - an open-fsync-close triplet, is not as bad as a 20second
regression on first-start from the bus-load of gross-ness that is done
in desktop's setup / migration code ;-)

> (*) I'm assuming here that this sync really matter only when saving
> 'documents'... and that we could live without it for other write...

        Sure - the only problem is is trying to push the semantic: 'we want to
fsync' through the multiple layers of abstractions ;-) Stephan - for
example, can we inherit UNO structs and be sure the new version is
correctly co-erced into the old one in other places ? [ that might be
helpful ]

        ATB,

                Michael

-- 
michael.meeks@suse.com  <><, Pseudo Engineer, itinerant idiot