Comanche - Failed primitives in Socket objects

Frantisek Fuka fuxoft at gmail.com
Tue Jul 13 16:55:47 UTC 2004


Thank you very much for the reply. II like Smaltalk very much but I
don't have much experience with Comanche and Squeak. (I did my
previous work in WisualWorks but I'd like to use something open-source
now.) Do you think I can use Comanche and Squeak to write a server for
real-world use that provides dynamic content without any other
software (Apache etc...)?

I am not a troll, I just honestly don't know and I'd ve very unhappy,
if I found out, after debugging the server at home, that it has
problems serving moderately busy real-world site.

Another question: Do I understand correctly that Comanche request are
not "forked"and are handled one-after-another? E.g. while I am
accessing my database to create a Response for one request, I can be
certain that another Request cannot mess up my database while I am
doing this? (Because this is what i want.)

On Tue, 13 Jul 2004 10:43:14 +0200, Kamil Kukura <kamk at volny.cz> wrote:
> Frantisek Fuka wrote:
> > While running the latest Comanche on Windows 98 and setting mode: to
> > #debug, I get lots of "Failed primitive" exceptions in Socket objects
> > almost immediately when I try a simple "load test" (holding down F5
> > key in my browser).
> 
> I sent to my HttpService #setDebugMode and tried it in Windows XP. I
> encountered same problem - that is primitive failed in
> Socket>>primSocket:sendData:startIndex:count:
> 
> According to this post:
> http://article.gmane.org/gmane.comp.lang.smalltalk.squeak.seaside/1228/
> I tried "TcpListener backlogSize: 100" and it seemed it cured the problem.
> 
> > Is this the result of limited number of sockets in Windows 98? I tried
> > similar test with Twisted Python frameword and there didn't seem to be
> > any problems. When running the server under Linux, in standard mode
> > (not #debug) will there be problems like this if the traffc is high?
> 
> Someone on Linux could try this "simple load test", though it's also a
> test for browser's event dispatch mechanism as well :)
> 
> > If the mode is not #debug, are all the exceptions correctly discarded?
> 
> It seems it might be the case; check comment in
> HttpAdaptor>>writeResponse:for:
> 
> --
> Kamil
> 


-- 
Frantisek Fuka (yes, that IS my real name)
(and it's pronounced "Fran-tjee-shek Foo-kah")
My E-Mail: fuka at fuxoft.cz
My Homepage: http://www.fuxoft.cz
My ICQ UIN: 2745855
FFFILM: http://surf.to/fffilm



More information about the Squeak-dev mailing list