mod_rivet experimental development status

[Massimo Manghi <mx...@apache.org>]

Hi everybody

I exploited the last days of vacation to do some more development on
the experimental code of mod_rivet. I want to thank George Petasis who
helped me a lot. George spotted several issues in my code and among
them he found the problem that broke mod_rivet when running concurrent
requests. He also noticed a problem when shutting down the server that
I found out was related with the way APR calls clean-up callbacks of
thread private memory. This forces us to adopt the bridge module in
rivet_worker_mpm.c whereas I hoped the much simpler
rivet_aprthread_mpm.c bridge could work. I'm considering to file a
feature request to the apr-dev list demanding to optionally control
thread private data management so that memory is freed synchronously
in the same thread that allocated it.

The aircraft can take-off, fly and land safely, but it cannot be
customized for your payload. Configuration of mod_rivet must be
rewritten so that threads don't share the same Tcl_Obj instances
storing configuration scripts. There are quite a few issues of
optimization as Tcl code will be byte-compiled several times, but I
think we can handle it.

Anyone with a working Windows development environment is invited to
step forward and join us in the efforts to create a bridge for the
windows MPMs.

  -- Massimo

---------------------------------------------------------------------
To unsubscribe, e-mail: rivet-dev-unsubscribe@tcl.apache.org
For additional commands, e-mail: rivet-dev-help@tcl.apache.org