On Wed, Oct 17, 2012 at 02:46:34PM +0200, Ruben Smits wrote:
> On Wed, Oct 17, 2012 at 2:40 PM, Markus Klotzbuecher
> <markus [dot] klotzbuecher [..] ...> wrote:
> > On Wed, Oct 17, 2012 at 11:13:14AM +0000, Ruben Smits wrote:
> >>
> >> http://bugs.orocos.org/show_bug.cgi?id=998
> >>
> >> Summary: Cannot self-update a LuaComponent
> >> Product: Toolchain
> >> Version: master
> >> Platform: All
> >> OS/Version: All
> >> Status: NEW
> >> Severity: major
> >> Priority: P3
> >> Component: OCL
> >> AssignedTo: orocos-dev [..] ...
> >> ReportedBy: ruben [dot] smits [..] ...
> >> CC: orocos-dev [..] ...
> >> Estimated Hours: 0.0
> >>
> >>
> >> Created attachment 830
> >> --> http://bugs.orocos.org/attachment.cgi?id=830
> >> testcase
> >>
> >> This bug is somewhat related to Bug 945: Can't self-configure/start a
> >> LuaComponent.
> >>
> >> Although self-configuring and self-starting is working, updateHook seems to be
> >> blocked:
> >>
> >> I tested by executing the attached testcase
> >>
> >> Executing this results in:
> >>
> >> rttlua -i test.lua
> >> Real-time memory: 517904 bytes free of 524288 allocated.
> >> OROCOS RTTLua 1.0-beta5 / Lua 5.1.4 (gnulinux)
> >> configure
> >> start
> >> >
> >>
> >> If I exit rttlua (Ctrl-D) the following is printed:
> >>
> >> update
> >> [...]
> >> update
> >> TLSF bytes allocated=524288 overhead=6384 max-used=6384 currently-used=6384
> >> still-allocated=0
> >>
> >> I think this might be caused by updateHook waiting for the mutex that is locked
> >> by lua_repl, but is blocked since updateHook is executed in the Activity thread
> >> which is not necessarily the same as the rttlua thread?
> >
> > Yes, the repl holding the mutex is running in the main thread, but the
> > activity is in a different one and has to wait. If you try to do this
> > in a separate component this works as expected, see attached modified
> > example.
>
> In the meantime have worked out something similar.
>
> > IMHO this is not a bug, this is misuse of the rttlua deployer :-)
>
> I understand how this is misuse, but could we somehow alarm the user
> about this misuse? Documentation, at runtime?

Maybe we should avoid it altogether. I'm wondering why the
rttlua-deployer needs to be a component at all. Instead we could just
run the repl from ORO_main and add a function rtt.getDeployer() or
similar. Or am I missing a good reason for this?

Markus