[MIR] speech-dispatcher

[Luke Yelavich]

 affects ubuntu/speech-dispatcher
 subscribe ubuntu-mir

Requesting that the speech-dispatcher package be promoted to main, as required by the gnome-speech replacement specification: https://blueprints.edge.launchpad.net/ubuntu/+spec/desktop-karmic-gnome-speech-replacement

Report URL: https://wiki.ubuntu.com/MainInclusionReportSpeechDispatcher

Thanks.

[Martin Pitt (pitti)]

Luke, please subscribe to speech-dispatcher bug reports (package bug contact), since this will probably need a fair amount of maintenance and cause some bug reports after karmic's release.

Init script shouldn't run in levels 0 and 6. (update-rc.d and LSB header)

postrm removes system user on remove, which is usually a bad idea, especially if there are any files owned by "speech-dispatcher". If at all, it should only happen on purge, but it's best to drop it completely.

Lots of patches that piled up; you said that upstream was dead, did you manage to get upstream commit rights, so that you can do changes directly upstream? Otherwise this will quickly get difficult to manage?

Nitpick: (not a MIR blocker)
W: speech-dispatcher: postinst-has-useless-call-to-ldconfig
Usually caused by dh_makeshlibs being too eager and picking up stuff in /usr/lib/speech-dispatcher. This usually needs an "DEB_DH_MAKESHLIBS_ARGS = -X/usr/lib/speech-dispatcher".

Otherwise this looks ok. It's a significant maintenance issue, but we are committed to it.

Can gnome-speech go to universe when this gets promoted? Please change the seeds accordingly already.

[Luke Yelavich (themuso)]

On Thu, Aug 27, 2009 at 05:08:37AM EST, Martin Pitt wrote:
> Luke, please subscribe to speech-dispatcher bug reports (package bug
> contact), since this will probably need a fair amount of maintenance and
> cause some bug reports after karmic's release.

Done, I thought the accessibility team was also subscribed, but better that I am in person as well I guess.

> Init script shouldn't run in levels 0 and 6. (update-rc.d and LSB
> header)

Ok thanks, will fix.

> postrm removes system user on remove, which is usually a bad idea,
> especially if there are any files owned by "speech-dispatcher". If at
> all, it should only happen on purge, but it's best to drop it
> completely.

Ok, that was from the debian package, but will remove it since our diff is rather big anyway, and I keep an eye on upstream closely, not that they are doing much these days. :(

> Lots of patches that piled up; you said that upstream was dead, did you
> manage to get upstream commit rights, so that you can do changes
> directly upstream? Otherwise this will quickly get difficult to manage?

I had cvs commit rights, but they moved to git, and so far, no community contributors who had cvs access have git access. A few of us in the community are pushing for upstream to do something about this, and even make a release tarball with all the work that has piled up since late last year, but we haven't heard from them yet.

I have a git tree of my own where these patches are committed, but since I don't officially speak for the project, I can't make a tarball available officially.
>
> Nitpick: (not a MIR blocker)
> W: speech-dispatcher: postinst-has-useless-call-to-ldconfig
> Usually caused by dh_makeshlibs being too eager and picking up stuff in /usr/lib/speech-dispatcher. This usually needs an "DEB_DH_MAKESHLIBS_ARGS = -X/usr/lib/speech-dispatcher".

Ok. This actually is something that needs to be solved in a better way, as the speech-dispatcher-flite package depends on a library in usr/lib/speech-dispatcher. I am having thoughts of dropping flite from main and not building support for it at all, which would solve this more cleanly.

> Otherwise this looks ok. It's a significant maintenance issue, but we
> are committed to it.

Maintenance which I am very willing to take on, due to pushing upstrea with the linux foundation a11y group for a speech API standard. There will be much more work going into speech-dispatcher in the coming 6 months, and it will likely be myself and a few others who will do the work, with upstream not likely to also chip in and help. If this ends up being forked, then myself and others will maintain it in our own time which should lessen the amount of work I have to do on it in work time.

> Can gnome-speech go to universe when this gets promoted? Please change
> the seeds accordingly already.

Yes, gnome-speech should be able to go to universe without issue, now that gnome-orca doesn't depend on it, as of latest upload.

Thanks

Luke

[Steve Kowalik (stevenk)]

I have promoted speech-dispatcher (source and binary), python-speechd, libspeechd2, and libspeechd-dev from universe to main, per Martin's approval, and noting that Luke has fixed the issues that Martin has pointed out.