You can find it either in the CVS on
http://sourceforge.net/projects/portableaserve/
You can download a tarball with the actual release at
http://www.dataheaven.de
Regards,
Jochen Schmidt
--
http://www.dataheaven.de
From: Andrew K. Wolven
Subject: Re: ANN: Portable AllegroServe 1.2.3a available (for LispWorks and CMUCL)
Date:
Message-ID: <3B7002C8.38C8C72C@redfernlane.org>
Jochen Schmidt wrote:
> You can find it either in the CVS on
>
> http://sourceforge.net/projects/portableaserve/
>
> You can download a tarball with the actual release at
> http://www.dataheaven.de
Thanks for your work in this area.
So I gather from reading this web page that your CMUCL port is a
single-threaded port?
Do you happen to know how it might perform on CMUCL x86 linux with
preliminary thread support?
AKW
>
>
> Regards,
> Jochen Schmidt
>
> --
> http://www.dataheaven.de
Andrew K. Wolven wrote:
>
>
> Jochen Schmidt wrote:
>
>> You can find it either in the CVS on
>>
>> http://sourceforge.net/projects/portableaserve/
>>
>> You can download a tarball with the actual release at
>> http://www.dataheaven.de
>
> Thanks for your work in this area.
>
> So I gather from reading this web page that your CMUCL port is a
> single-threaded port?
> Do you happen to know how it might perform on CMUCL x86 linux with
> preliminary thread support?
No it uses the stack-group based threadsupport of CMUCL on x86.
Maybe it would be a good idea to add a serve events version too...
Another possibility is the use of forks and therefore real UNIX-Processes.
AllegroServe supports this under AllegroCL/UNIX but it is untested under
LispWorks and CMUCL. The functionality is in and all needed
foreign-functions (like fork() ) are defined but as I said this is
untested...
I will give it a try and will send you a message if this works.
ciao,
Jochen
--
http://www.dataheaven.de
From: Bulent Murtezaoglu
Subject: Re: ANN: Portable AllegroServe 1.2.3a available (for LispWorks and CMUCL)
Date:
Message-ID: <87y9ovyj5y.fsf@nkapi.internal>
>>>>> "JS" == Jochen Schmidt <ยทยทยท@dataheaven.de> writes:
[...]
JS> Another possibility is the use of forks and therefore real
JS> UNIX-Processes. AllegroServe supports this under
JS> AllegroCL/UNIX but it is untested under LispWorks and
JS> CMUCL. The functionality is in and all needed
JS> foreign-functions (like fork() ) are defined but as I said
JS> this is untested...
Not completely on topic, but this got me thinking about whether one
should invoke the GC before calling fork. I am thinking of the case
where lots of short-life objects are created by a program before forking
several more processes. This doesn't hurt anything initially because of
copy-on-write, but the subsequent GC becomes expensive for all the
processes and needlessly so IMHO. Just an intuition, but I'd be
interested to hear what others' experiences are.
cheers,
BM