On 05-06-2015 11:44, Paolo Bevilacqua wrote:
Yes, it's clear. Note in the above example one ends with multiple
concurrent log files, witch is not a common approach for logging.
And, must consider that using database or even IPC for persistence
comes
with a price, mainly in terms of complexity and learning curve.
Both things that I can't afford in the current project. TCL generally
is
great in terms of simplicity and immediate power, but less than that
in
this specific case, I believe.
Paolo, your remarks must not be directed to Tcl, but to Apache. This is
the architecture of the webserver and we embed Tcl in it, we don't use
Apache to run Tcl. It looks like your problem is better addressed
writing a pure Tcl server or perhaps using tclhttpd.
A crucial point of the Apache HTTP webserver is a complete agnosticism
about the underlying model of multiprocessing (in fact several are
supported). Otherwise you wouldn't have the extreme modularity the
Apache HTTP Webserver has. Therefore having one process, 10 processess
or 10 processess running 10 threads each shouldn't make any difference
and therefore any communication among this agents has to be addreessed
with specific methods at application level. I have had for long time in
my TODO list the project of implementing such a IPC scheme with a Tcl
interface based on shared memory, but never had time (or urgence) to
work on it
ciao
-- Massimo
P.S. Please, in order to amend our documentation would you tell us what
is the page of the manual you found misleading?
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
