On 2016-02-28 10:36:25 +0100 Richard Frith-Macdonald <richard.frith-macdon...@brainstorm.co.uk> wrote:


On 28 Feb 2016, at 09:20, Riccardo Mottola <riccardo.mott...@libero.it> wrote:

Hi,

I experience lately problems with "someone broke our lock" issues and other Lock related exceptions. I have experienced this on NetBSD, OpenBSD and Linux, clang or gcc, it is not OS nor compiler dependant. It happens on plain uniprocessor computers.

I know that Gregory had some issues too. Greg, maybe you can add information?

Usually the nuiseance happens when launching a GS application when one is already running (e.g. one runs Terminal or GWorkspace and launches application X).
I think that latley it is happening really a lot.
The most common thing is that you canot start an application and get a popup. Retrying a couple of time helps. Sometimes, like what happened just to me, you leave things running and get back to the computer and found one app died leaving that in the console.

Maybe writing defaults is related, but it could be something related.

It surely makes using more apps and thus having a Desktop quite have a bad thing now. I think we need to dig it out before release. The issue is... I have no way to reliably reproduce that

It sounds like you are probably talking about filesystem locks (inter-process) rather than processor (inter-thread) locks. NSUserDefaults uses NSDistributedLock to ensure that only one process updates the persistent defaults at a time, so you could try looking there. If I'm to look into it under debug, I really do need at least a way to reproduce the problem though.

I started GWorkspace and saw this:

/System/Tools/fswatcher: Uncaught exception NSGenericException, reason: Unable to get attributes of lock file we made

Then I started GNUMail (to report hits error) and saw this:

Warning ... someone broke our lock (/home/multix/GNUstep/Defaults/.lck/.GNUstepDefaults.lck) ... and may have interfered with updating defaults data in file.201

Does this give you any ideas? This is on NetBSD/x86/clang it happens often, but not always...


Riccardo


_______________________________________________
Gnustep-dev mailing list
Gnustep-dev@gnu.org
https://lists.gnu.org/mailman/listinfo/gnustep-dev

Reply via email to