On 3/1/2016 6:31 AM, Pedro Giffuni wrote:
Hi Patricia;
Good analysis, thank you!
I think the operation is invalid. We can just comment out the code.
FWIW, I looked at the SUN history to see if I could guess what the
author pretended:
https://bitbucket.org/mst/ooo340/annotate/8b564af029aa60259c0d133d7dc37468ef56e4b5/sal/osl/unx/profile.c?at=default&fileviewer=file-view-default
But it's not clear at all.
This is exactly what I meant when I said the code is ugly :-P.
I agree that the code is ugly. Dennis Hamilton and I are not even sure
whether it is currently used. See the header file profile.h.
I don't know whether the consequences of commenting out the unlock would
be better or worse than dereferencing a null pointer, if that code is
ever reached. In general, the longer a program goes on running in an
anomalous state, the greater the risk of file corruption.
My recommendation is that we do need to understand what is going on,
what the lock means, what a null pointer their means etc. before making
any change. Dennis Hamilton and I are working on that, but more eyes on
the code may be useful. As I suggested earlier in the thread, forming
small informal teams to work for a while on a particular issue may be
better than each of us working independently.
Patricia
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org
For additional commands, e-mail: dev-h...@openoffice.apache.org
