Re: Presentation + A debian-based for audio creation and production, stage technics and video blend (or "the future of TangoStudio")
Hi Aurelien,
thanks for your interest in Debian Pure Blends.
On Sun, Nov 11, 2012 at 03:14:13PM +0100, Aurelien wrote:
> [ cut long but interesting and informative introduction ]
>
> I've been searching for some days the solutions we might chose to build
> such a distribution, and after spending time on simple-cdd, live-build,
> or again remastersys, Charles Plessy finally told me to check Debian
> Pure Blends, and I must admit what I'm talking about is clearly a blend.
> After looking a bit, it seems to fit perfectly an existing blend, which
> is the Debian-Multimedia one, and from the discussion I had with some of
> you on IRC this morning, I think I'm not wrong.
+1
> So, here it is, I'd be interesting in knowing whether that project could
> enter that blend, and myself at the same occasion. It's a true question
> for me as I'm just discovering that blend stuff, and I don't want to do
> some mess in something already working, or whatever.
I really welcome your attempt and as the person who have done a lot of
Blends code (and with a lot of more code ideas in mind if time would
permit) I would really like to support your intention.
> There are also some people wanting to involve (semi-dev like me, or
> bêta-testers, or again people with communication/web skills, the ones who
> were involved in TangoStudio). From the discussions on IRC this morning,
> it seems that blends shouldn't be restricted to dev only, but could also
> be opened to "specialised users", so they could join the blend too, but
> once again, I don't want to do too much mess.
To my experience there is not necessarily a conflict between supporting
users and developers. In some Blends (Debian Med and Debian Science we
created tasks (Blends jargon - see doc[1]) targeting at users (for
instance med-bio targets at the user of bioinformatics packages to
analyse gene sequences etc) and other tasks for developers (for instance
med-bio-dev targeting of developers of bioinformatics applications). I
have no idea whether this idea can be applied in Multimedia field - just
mentioning that there is no point in restricting your focus on pure
multimedia users beforehand.
> To end up, we're currently listing all the features that were in
> TangoStudio and that appear to us as "essentials" for that type of
> audio/lights/video specialised users. The list is not complete by now,
> but clearly, it has to deal with:
>
> - having realtime property, priorities, memlock, open file numbers and
> so on preconfigured,
> - avoiding pulseaudio by default
Both items were just discussed
> - having several packages not in debian yet (Non quadrilogy, Tapeutape,
> Tranches, Q Light Controller...), and maybe having an experimental
> branch for bleeding-edge softwares to be tested ; also having several
> packages compiled with different options from Debian official (this
> might be discussed)
In addition to the very useful links Paul has given in his answer I
would like to mention that running a Blend inside Debian (if done with
some engagement) is a very clever way to effectively form a team that
cares for a set of packages and it becomes easier to get software
included straigt into Debian. I recently did a quick hack producing a
graph of the development of packages of certain main tasks in Debian
Med[2] and we also create graphs about the people working most active
in our teams (see for instance [3a] for Debian Med and [3b] for Debian
Multimedia).
BTW, I could write down some paragraphs about why it is important that
packages are included into official Debian - but I guess from your mail
that you are just aware about this. One of the main important things is
actually fixing the original problem why you finally ended up here: We
have seen *lots* of specialised distribution (not only in Multimedia)
that tried to address specific user interests and finally most of these
are stalled / unmaintained now because they never gained enough critical
mass to keep on the initial momentum. In Debian you have some critical
mass from the first moment and you can be quite sure that some "good"
software will remain in Debian (I'd even go so far that software that
was inside Debian and is not keept there is most probably not "good" -
otherwise somebody would rise up and save the software inside.)
> - a (graphical) installer with more options/questions than it has by now
> (like what GUIs do you want: XFCE, LXDE, TouchScreen GUI, none, or
> again what softwares pack should be installed at installation time:
> audio production, audio for live, video, base-system, and so on),
Answered by Paul.
> - a selection of preconfigured WM from the Light-Gnome that TangoStudio
> had, to a very light Openbox, without forgetting that touchscreen
> thing which is more and more used in live conditions (sound guys often
> can't make the difference between an OS and the WM they use)
I'd like to expand Paul's argument a bit. Providing some kind of strict
default WM might give some group of users a very safe feeling and they
will feel very comfortably home ... and other users just don't. They
will refuse to use this GUI (for whatever reason) and if they are bound
to your choice they will not use this Blends at all. IMHO Blends should
be orthogonal to GUI wars - just support user specific application and
leave all other free choices. Yes, the avialability of free choice is a
real problem for newcomers because they need to trust some
recommendation first to not make some stupid wild-guessing inbetween all
the options and another problem is that it is hard to create some kind
of "corporate design" which is probably a thing you want to do. Perhaps
you just try some experiments but please keep these pros and cons in
mind.
> and first of all, this "blend" or "distribution" or "Debian Custom"
> should be packed as iso files, so that anyone might install it easily
> (as TangoStudio did).
As Paul said there is no such general thing but Debian Edu does a lot of
stuff in this direction. Perhaps you might be able to steal /
generalise some stuff (so other Blends could profit from this as well).
> Well, here it is. It was long, sorry for the convenience, and
It was not too long but informative enough to (hopefully) give some
reasonable advise.
> congratulations if you read until there!
Thanks for the congratulations and here are comming further hint:
Currently the tasks of Debian Multimedia are maintained in SVN[4] (after
release of Wheezy they will be moved to Git as you can read in my
announcement[5].) To get write permissions (to both SVN and Git) you
need to become a member of Debian Pure Blends group on Alioth[6] (at
least currently it is organised like this, once we will have switched
to Git we might or might not move the tasks stuff also to the relevant
teams project.)
All these tasks are basically created by myself (a few commits are done
by real Debian Multimedia team members) which in turn means: They are
most probably quite useless because I'm simply no multimedia user and
have done probably choices which are simply not attractive. In other
words: Please do not regard anything as fixed but rather feel obliged
to change whatever you feel is broken. If you want to get a simple
overview about the tasks layout you can quickly browse the web sentinel
pages[7]. These pages are created in a daily cronjob out of the content
of the tasks and the tasks files are described inside the Debian Pure
Blends docs[8].
By using the package blends-dev in principle the tasks could be turned
into metapackages containing exactly the dependencies specified inside
the tasks. This was not yet done because as I said I have no idea
whether my crappy knowledge would turn out into reasonable metapackages.
My basic intend was to demonstrate to Debian Multimedia team what might
be possible and I was (unsuccessfully) hoping that some of its team
members would be excited by the idea and do something in this direction.
However, some work was done by Rosea Grammostola (in CC) who did some
discussion in the past (starting at [9]) about the very same things like
you are wondering now. He created some external Git archive[10] (which
was not touched since Thu Jul 28 22:50:26 2011 - see above for the lost
momentum effect when working outside Debian) and what I really recommend
you to look at. I always intended to ping Rosea what he might think
about merging his stuff into the currently available tasks and may be we
could finally profit from his work which looks way more knowledged and
reasonable than my crappy assembly of packages.
Apropos SVN to Git migration: If you would confess that you are a Git
addictive I might consider to fast-hack the tool that creates the web
sentinel to start the migration of non-metapackage building projects
right now before the release of Wheezy. This should not be a lot of
work and I would like to send the message that we are interested in your
plans and will support your ideas as much as possible.
Kind regards
Andreas.
[1] http://blends.alioth.debian.org/blends/
[2] http://anonscm.debian.org/gitweb/?p=blends/website.git;a=blob_plain;f=misc/team_analysis_tools/debian-med_1.12.pdf
[3a] http://debian-med.debian.net/
[3b] http://blends.alioth.debian.org/multimedia/
[4] svn://svn.debian.org/svn/blends/projects/multimedia/trunk
[5] https://lists.debian.org/debian-devel-announce/2012/10/msg00008.html
[6] http://alioth.debian.org
[7] http://blends.alioth.debian.org/multimedia/tasks
[8] http://blends.alioth.debian.org/blends/ch-sentinel.en.html#s-edittasksfiles
[9] https://lists.debian.org/debian-blends/2010/10/threads.html
[10] https://github.com/johnsen/meta-blends
--
http://fam-tille.de
Reply to: