Padraig +1
As you say we need to review the rest of the code to catch other
situations where we might run into this problem, tracked in:
8424 GUI should use error handler for out of memory conditions
JR
Shawn Walker wrote:
On Jun 24, 2009, at 3:32 AM, Padraig O'Briain wrote:
Shawn Walker wrote:
On Jun 23, 2009, at 12:22 PM, Padraig O'Briain wrote:
Shawn Walker wrote:
On Jun 23, 2009, at 9:16 AM, Padraig O'Briain wrote:
The webrev, http://cr.opensolaris.org/~padraig/ips-9501-v1/, fixes
9501 packagemanager traceback on memoryerror
The fix is to do catching memory errors when installing or
removing in the same way as the cli.
Are you sure this shouldn't be a general top-level exception case
in packagemanager.py instead of just for the installupdate case?
Of course I don't know if you can do that.
You might want to do this for updatemanager as well (another bug?)...
This change will work for updatemanager also if the memory error
occurs during update.
I need to think about whether we can do something in
packagemanager.py. Even if we do, I think that this webrev should
be applied.
the code in packagemanager/py is executed in
1) a thread
2) a callback launched by GTK event handler.
For 1) we could create our own thread invoker which catches Memory
errors
I do not see a general solution for 2). I tried pytting a top-level
handler around the packagemanager.py code and raising a MemoryError
in a callback. I find that it does not make its way through the C
code in GTK to my handler.
I would like to regard this as a separate issue to 9501 as most of
the memory issues which occur in packagemanager and updatemanager
will occur in installupdate.py.
I think that's fine, but that does mean you need to 'catch'
MemoryErrors in lots of places.
The more immediate need is definitely during install/update though.
Cheers,
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
