I'm new to puppy (and linux) but I believe that this is a really interesting discussion though I would think that may have been done before in other forums and distros.
I think that at the end of the day Linux development follows the "scratch an itch" approach,
your itch. So is up to the actual developers to do whatever they want. All the rest can do is to tingle them hopping for a scratch "at the right place", but not much more.
I think that the major issue is to decide what puppy is/will be. So far is whatever BK wants it to be. But what are the fundamental traits that make it a puppy. In development terms (small, fast, suitable for old/new hardware, user friendly etc does not qualify). That runs in RAM? The layered file system? The stripped apps/libraries? BK's scripting? The gtk-based UI? all of the above? other?
Then there is the issue of what puppy is going to be for the end user. Stable or bleeding edge? For old hardware or for 8-core computers? Should it be an OS that mostly runs from a CD/USB in RAM or as full HD install. Can all of the above be achieved in a 100-150MB OS or pupplets is the only way to go? Finally, is there going to be only a central approved repository of pets or the forum will remain a major source of pets?
I would be surprised if more that 3 people could agree in any given set of answers on these
. However the Lupu/quickpet is probably a good paradigm that could be coupled (metaphorically) with Slackware's hardware-detection abilities and Gentoo's hardware-specific installation, and get extended into a "remaster on first boot" system.
Eg a minimal system is booted (available in "old" and "new" kernel versions) that basically should be able to detect the hardware, run a VESA/TinyX/fbdev display driver and LAN-connect to the internet. Then hardware drivers can be downloaded as needed and applications can be downloaded on user demand or pre-configured blobs. Then the custom puppy can be "remastered" and saved to the indicated media for reboot.
The LAN connection could be replaced by a fairly bigger initial download that would include most of the commonly used drivers/apps and used in areas without a LAN connection.
See it like a "woof build system" that will run of the top of the minimal system with input from the hardware detection and the user preferences and re-master puppy of the fly
.
This approach allows developers to work in their area of expertise or preference without being forced to remaster and maintain a pupplet. Would allow both stable and bleeding edge pets to co-exist and leave it to the final user to decide which way to go. Would allow 486 and 8-cores to be supported (probably through different kernels). Would allow puppy to run as OS-on-a-stick or a server etc.
It would also probably produce smaller and probably faster puppies without the extra baggage for other hardware or the 5 calculators
However, (besides possible insurmountable technical difficulties) there is the ease-of-use and out-of-the-box experience, that will suffer from such an approach. But nobody stops anyone in developing pupplets... if anything will make it even easier!
On the other hand no-matter what people say, "linux newbe" is still an oxymoron. You either get comfortable with the command line or you get out of linux fast...
I'm clearly not claiming that I have the solution for the issues at hand, all I'm saying is that to accommodate all these contradicting needs/views/priorities and bring people to work together you need a highly customizable OS that will fulfill both developer and user priorities/needs.
Of course the planets could align correctly and a significant number of developers or even all, agree on a given puppy. It is possible I just do not think is vary likely (and this is NOT an accusation to any extend).
Just my 2 (newbe) cents.
== [url=http://www.catb.org/esr/faqs/smart-questions.html]Here is how to solve your[/url] [url=https://www.chiark.greenend.org.uk/~sgtatham/bugs.html]Linux problems fast[/url] ==