Re: Testing, systemd support, rfc5188, cleanups

On Tue, Feb 14, 2012, at 20:32, Will Manley wrote:
> On Mon, Feb 13, 2012, at 23:16, Juliusz Chroboczek wrote:
> > >  * automated-testing introduces systemd socket activation support
> > 
> > No way, sorry.  Polipo can be reliably restarted by systemd (and runit,
> > and daemontools, and V7 init) without the need to include 600 lines of
> > code that I don't understand.
> 
> For clarity, systemd support was added for socket activation rather than
> process supervision.  Indeed we don't actually use systemd here, but we
> require the ability to pass a socket to polipo so we can create a
> private sandboxed HTTP cache, per application, on our embedded set-top
> Linux without having to wait for polipo to start up to query it for
> which port it has opened on.  We chose to make it systemd compatible as
> we thought that that might be more useful to others.
> 
> I agree that adding 600 lines of sd_daemon.[ch] is a little bit silly
> considering that the bits that are used are just checking that
> getenv("LISTEN_FDS") == "1" and getenv("LISTEN_PID") = getpid().  I will
> rewrite the patch set to not include those files.

I created a new patch set adding systemd support and the test framework
but avoiding adding systemd's sd-daemon.[ch] (the aformentioned 600
lines of ununderstandable code).  I'd like to reiterate that systemd
support is not my goal, but socket passing is so I thought I'd at least
make it systemd compatible.

Please take a look and reconsider.

It can be found with:

    git pull git://github.com/wmanley/polipo.git automated-testing2

or at

    https://github.com/wmanley/polipo/tree/automated-testing2

Thanks

Will

------------------------------------------------------------------------------
Virtualization & Cloud Management Using Capacity Planning
Cloud computing makes use of virtualization - but cloud computing 
also focuses on allowing computing to be delivered as a service.
http://www.accelacomm.com/jaw/sfnl/114/51521223/