On Feb 27, 2009, at 3:56 AM, Leslie P. Polzer wrote:
>
>> LPP> I spawned a thread that does some heavy slot access. So maybe
>> LPP> the serializer isn't entirely multi-thread or multi-process safe
>> LPP> yet?
>>
>> maybe. there were some issues with it (trivial ones), who knows,
>> maybe we've
??>> if you have a multi-core system and value stability over performance,
??>> i strongly recommend binding you lisp process to a single core (via
??>> taskset in Linux, for example). this way multithreading-related issues
??>> will have much less chances to happen.
LPP> No choice in a web s
> LPP> I spawned a thread that does some heavy slot access. So maybe
> LPP> the serializer isn't entirely multi-thread or multi-process safe
> LPP> yet?
>
> maybe. there were some issues with it (trivial ones), who knows,
> maybe we've overlooked something.
We should probably consider adding a
> I thought we resolved those issues some time back (e.g. accidentally
> sharing a buffer-stream). It's possible there was a regression, but
> nothing obvious comes to mind. I presume this same thread operated
> fine prior to your upgrade to alpha?
No, not really. We've had serialization errors
LPP> I spawned a thread that does some heavy slot access. So maybe
LPP> the serializer isn't entirely multi-thread or multi-process safe
LPP> yet?
maybe. there were some issues with it (trivial ones), who knows,
maybe we've overlooked something.
if you have a multi-core system and value stabil
PS - I do have a live system that I've been incrementally migrating
since September without any upgrades or fixes, so I don't think there
is an obvious incompatibility.
Are you on a 64-bit or 32-bit machine? (and SBCL/Linux, I presume) I
cleaned up some optimization code in memutils relatin
I thought we resolved those issues some time back (e.g. accidentally
sharing a buffer-stream). It's possible there was a regression, but
nothing obvious comes to mind. I presume this same thread operated
fine prior to your upgrade to alpha?
Can you generate a diff for me? I might catch so
> Can you give more information about the condition under which it
> cropped up?
I spawned a thread that does some heavy slot access. So maybe
the serializer isn't entirely multi-thread or multi-process safe
yet?
Leslie
--
LinkedIn Profile: http://www.linkedin.com/in/polzer
Xing Profile: htt
Can you give more information about the condition under which it
cropped up? -Ian
On Feb 24, 2009, at 6:28 AM, Leslie P. Polzer wrote:
>
>>I provided a patch to fix the problem occured in the 0.9.1->1.02a
>> upgrade process.
>>The issue occurs when doing migrate with SBCL. The deseriali
> I provided a patch to fix the problem occured in the 0.9.1->1.02a
> upgrade process.
> The issue occurs when doing migrate with SBCL. The deserialization
> exception is reported.
> The latest email chain includes more details.
Thanks, but I wasn't using MIGRATE at all...
--
Linked
Hi Lesile,
I provided a patch to fix the problem occured in the 0.9.1->1.02a
upgrade process.
The issue occurs when doing migrate with SBCL. The deserialization
exception is reported.
The latest email chain includes more details.
If the failure symtom is similar, you may want to u
11 matches
Mail list logo
