What would be the advantage of starting a distribution from scratch?

On the negative side I see that meego will loose on some areas:

1) The amount of knowledge that can be harvested from the existing forums,wikis. Using, let's say, distribution X as starting point, one can figure out what to do to fix or adapt things in a new derived distribution.

2) Tested and packaged software (specially, ARMel and all others that are not x86 compatible). yes, sometimes you must recompile and repackage, but it is easier when there is already a working software that expects more or less the same libraries to work.

3) As it was already pointed out on the dev-list at meego, if a new distribution follows closely the structure of a base one, it makes it easier to just use prepackaged software as much as possible. It might be the case of automatically rebuilding common packages, for instance. Now, since it is a new distribution, it will be harder to do that, because things will probably be placed on different places with different configuration options and so on.

I have an opposite experience. Never had a problem with YaST but constant problems with apt/dpkg.

I don't argue here - if we have a better GUI - use it. But I don't know any GUI installer on N900 beyond HAM. Synaptic? I don't see it's port to N900...

And the whole point of my previous post was:

Me vote for this too

For me .deb and .rpm are the same thing, a glorified tar-ball.
Sure i prefer rpm because I've built so many of the damned things, but it only took me about 15minutes to build my first .deb
If you really want to you can use deb on fedora and probably rpm on debian (shudders at the thought of either!)

some of their arguments are:

* who needs a large software pool? who needs "bind" on a smartphone?
* to build highly optimized packages, not desktop oriented ones
* straight ports without UI adaptation are not finger-friendly and therefore useless
* to strip out unneeded components
* less packages means less problems/bugs,
leaving out packages cuts a lot of security work
* ARM fixes should go to upstream anyway
* MeeGo has lots of paid developers who could do the porting
* some small distributions are much more efficient than the large ones
* being compatible involves a lot of work
* Maemo users and developers will adapt anyway
* Moblin has been doing it its own way and will continue to do so
* who cares?

Let me summaries my position:

RPM vs DEB - I read the whole thread and I want to say that it is not packaging format but the paradigm behind of it.

DEB allows many tricks ==> developers use it ==> end customer has problems with it.

RPM more weak at tricky points and developers are required to follow it and because of that some debian-based problems disappears. For exam, because less sophisticated dependency management the LIBRARIES are distributed with more attention and there is less problem with interdependencies. Nobody would create a situation like existence of libqt4-core with libqt4-maemo-core which (I believe) are mostly some derivate versions.


On other side - GUI vs command line tools:

I used 'configure/make/make install' many years. Couple of years I was forced to use YaST2 (because of SuSE - changed a workplace). I never had any problem with YaST2, and I noted that dependency handling in YaST2 is pretty user-friendly. It lists all packages which it will install in advance and in friendly manner you can chose that to do. The same is for conflicts and you have a good chance to explore dependency.

Command line tools never give that kind of pleasure.


Finally, specific words about apt/dpkg packages. Last time then I saw that kind of half-cooked applications was around 15years ago. I was choked then I got messages about DELETING of some packages then I just tried to DOWNLOAD it from repository. And it also tries ask my confirmation about it! (and silently and happily just forget about positive answer after completing download... but you never sure until it finishes)

And this is not only one - try to construct a command to download only ONE package from repository. Try to download a SPECIFIC version of some package having some newer version (yes, it is possible but it reminds me an ASSEMBLER programming... I know, I know, the real programmers never use Pascal...)

At this point I'd like to see some long term plans and commitments...

Let's stop these constant REVOLUTIONS and stick to some OS, ui toolkit, etc and start the EVOLUTION!

The libqt4-(maemo5) dependency fiasco has nothing to do with apt, deb or sorts (deb based distros have zero problems upgrading from 4.5 to 4.6). In Maemo's case it's just a hack that is there to satisfy certain Nokia policies. If it was RPM, it would be just as bad.

Otherwise, you're just saying you're more used to RPM based tools, nothing inherently inferior (we have synaptic if you're afraid of the command line, and appending =version on the end is not exactly rocekt science either

That's, unfortunately, not the Linux way. Rather than EVOLUTION the answer is often DISTRIBUTION. Then when two groups have to merge, all divergent choices need to be resolved, not always in the way that all parties would want.

Don't think of it as an obstacle, think of it as an opportunity. At least that's what the inspirational poster I see at work says. Yeah, I don't believe it either.

Yes and no: in case of RPM there is a chance that Nokia choses a different policy. Because of restrictive RPMs

Sorry, but (https://garage.maemo.org/projects/synaptic/) -

"This project has not released any files."

Let's go back to ground.

But wait:
https://garage.maemo.org/projects/yast2

Invalid Project

Ahh darn. No GUI for RPM either?