new target: media

Stefan Sperling stsp at stsp.in-berlin.de
Mon Aug 28 14:39:21 CEST 2006


On Mon, Aug 28, 2006 at 02:02:53PM +0200, Amadeus wrote:
> Hello everyone,
> 
> I am thinking about creating a new target in CVS, a common distribution 
> for all RAM-based cards with SD or CF cards. This distribution will 
> contain drivers for all cards, and will be loaded with applications.

> There is ongoing demand for a "stable" vendor from the DSLinux users.

This demand has been there all the time - dslinux has been in too much
of flux to produce a stable release yet. Hence we don't even have version
numbers.

I'm all for a stable release, but keep in mind that maintaining
a stable release alongside development branches takes time as well.
There will always be bugs or other issues in a stable release, no
matter how long you wait before releasing it.

So we should try to create a well working system for a stable release.
We don't want to go through release engineering (quite a huge task
as I currently experience at work) to release something that is
obsolete after 3 weeks - or nobody will use it.
A release should be "top-notch" for at least half a year, otherwise
it's not worth the effort. At the current rate of development,
builds get obsoleted in a matter of days.

Also, I see other (small) issues that need to be resolved before a
stable release, aside from the RAM issue (which I would still flag as
highly experimental at this stage by the way).

They are all on http://www.dslinux.org/wiki/TODO, including:
* Get ssh server working.
* Add multi user support by making /etc/passwd writable at runtime
* Fix various bugs that prevent bsdgames from running properly.
  Move games from /usr/bin to /usr/games. 

There are more points on the list, such as port email tools etc.
However, for a stable release I would suggest that we stop focussing
on adding functionality and instead fix things that are already there
but not working properly at the moment.

A stable release should also support a lot of different add-on
carts - creating a stable release for one build type only is not
worth it.

Also, we should aim for a stable release of the toolchain before
aiming for a stable release of dslinux. As issues with the toolchain
still keep popping up, I don't see a stable toolchain release really
soon (probably in a couple of weeks). And we have no need to rush.

> So SCSD-RAM will be a development vendor, and can have fast turn around 
> cycles, and experimental features, which are later enabled in the 
> common vendor.

I'm still in favour of renaming SCSD-RAM as well.

I don't think it is reasonable to create a "stable" SCSD-RAM
build alongside an experimental one, given the criteria for
a "stable" release I've outlined above. Users can (and sometimes
have to) wait to get the quality they demand.

There's also nothing stopping people from getting involved as well
if they want things get done earlier.
And there's nothing to gain for us by trying to get things done for
others as fast as possible. Where's the fun in that?

> Thinking about a name for this vendor, "MEDIA" comes to mind.

What about "RAM"?

PS: Let's decide on the name before creating the directory this time :)
PPS: The "vendor" would be us developers - what you mean is a "build type".

-- 
stefan
http://stsp.in-berlin.de                                 PGP Key: 0xF59D25F0



More information about the dslinux-devel mailing list