[Tts-project] Advice/opinions asked about two speech-dispatcher issues

[Samuel Thibault]

Hello,

Paul Gevers, le Sat 18 Jan 2014 22:36:39 +0100, a écrit :
> On 15-01-14 20:51, Paul Gevers wrote:
> > Second, I am unsure if speech-dispatcher upgrades are actually handled
> > gracefully. If speech-dispatcher is already running at the start of the
> > upgrade, I think it will keep on running and only the new version will
> > be used after killing speech-dispatcher (via reboot). However,
> > implementations that interact with speech-dispatcher might be upgraded
> > at the same time and actually believe they are talking to the new
> > speech-dispatcher and possibly fail. I can imagine that we could kill
> > any speech-dispatcher that is running in the postinst script, but I am
> > reluctant to take that step. What do you think?

Yes, the brltty package avoids restarting the daemon on upgrade.

> I think I will add a debconf question to the package, that asks if
> speech-dispatcher should be killed to allow restarting. Or do you think
> this would be overkill?

I would have rather thought about a notice to the user that some
applications may not be able to connect to speech dispatcher until the
daemon is restarted, which can be done manually or by just rebooting.
The debconf question seems overkill to me.

Samuel