Hang/crash in templates

[Lionel Elie Mamane]

On Thu, Apr 25, 2013 at 01:29:00PM +0200, Miklos Vajna wrote:
> On Thu, Apr 25, 2013 at 01:08:47PM +0200, Lionel Elie Mamane <lionel at mamane.lu> wrote:

>> If you can point valgrind/gdb it yourself and point me towards the
>> place / stacktrace that allocates "a lot", I can look at that place
>> and see if I get inspiration (before my build finishes).

> On the other hand, for now I would just change back the cache format.
> Does that make sense?

I don't know what cache you are talking about. If some binary (as
opposed to XML) format got changed incompatibly, then yes, this is
bad.

> I mean if you are OK with pushing
> http://cgit.freedesktop.org/~vmiklos/lo-core/commit/?id=83544b27f3dcfb132b3eb46d387f62ebc9c3e160
> to master.

To stay backwards-compatible in *values* (and not only in length of
fields in binary format), you will need:

#include <tools/time.hxx>

SvStream& operator << ( SvStream& _rStorage, const util::DateTime&_rDate )
{
    _rStorage << sal_uInt16(_rDate.NanoSeconds / ::Time::nanoPerCenti); // sal_uInt32 -> sal_uInt16

and
SvStream& operator >> ( SvStream& _rStorage, util::DateTime& _rDate )
{
    (...)
    _rDate.NanoSeconds = nTmp * ::Time::nanoPerCenti


-- 
Lionel