On Wed, 2012-03-28 at 10:04 +0200, Cor Nouws wrote:
> Indeed ;-) See issue fdo#47983
Which comes down to combination of problems:
* of not easily being able to push more information into the
file ucb to be able to flag that we want to call osl_syncFile
at just-this-one-place (saving a document), due to some
"shove arguments into a struct into an any" design - or
perhaps I'm missing something.
* the desktop/ migration code heavily using UCB.
* the desktop/ migration code being write-happy, and
re-writing the same couple of files 40 times or so
each ;-) [ and this is before the berkely database
cruft ]
So - unclear how to deal with that really. A very quick, very ugly hack
would be to stop the file ucp doing it's (new) sync if the path is
in /tmp or in your configuration ( I suppose ). Not at all beautiful,
but functional.
Failing that, we could re-open the file, purely in order to 'fsync' it
(conceptually yucky too) in SfxMedium::TransactedTrasferForFS_Impl and
pull the fsync out of the UCB's write (where it belongs). Or - we could
do something nastier ;-)
Thoughts Stephan ?
Hmm,
Michael.
--
michael.me...@suse.com <><, Pseudo Engineer, itinerant idiot
_______________________________________________
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice
