Puppy 6

Under development: PCMCIA, wireless, etc.
Message
Author
User avatar
tronkel
Posts: 1116
Joined: Fri 30 Sep 2005, 11:27
Location: Vienna Austria
Contact:

#31 Post by tronkel »

ttuuxxx wrote:
I look at it this way if we can shave 5-10MB+/- off a iso why not?
Take it further!

Why not do something similar to Arch Linux and only provide a core system to which you add only the stuff you want/need? The core doesn't even have a desktop. Arch Linux has done extremely well with this approach - but it's main afficionados are the techie types.

In the case of Puppy though, adding in the extra stuff should be made easier than it would be with Arch. The main desktop and network modules would be provided via the internet in the form of sfs modules - pre-configured and ready to go - just plug them in. Same for the browser.

Each developer would be in charge of his own module which would plug in to the same core as the other developers.

So that's five modules that would need a developer apiece:
Core, Network, Desktop, Video and Browser.

Other modules would be necessary as well, but these five should take care of a minimum working installation.
Life is too short to spend it in front of a computer

User avatar
greengeek
Posts: 5789
Joined: Tue 20 Jul 2010, 09:34
Location: Republic of Novo Zelande

#32 Post by greengeek »

tronkel wrote: So that's five modules that would need a developer apiece:
Core, Network, Desktop, Video and Browser.
Are drivers always part of the core? I have always been confused about what zdrv is. Is it possible to have a small, stable, tight core that is capable of automatically "pulling in" drivers (or requesting what it doesn't already have..) on a system that has new versions of hardware?

If the "driver section" was another one of the modules it could make hardware compatibility and portability easier perhaps?

User avatar
greengeek
Posts: 5789
Joined: Tue 20 Jul 2010, 09:34
Location: Republic of Novo Zelande

#33 Post by greengeek »

.
Puppy 6 and modularity? Great idea! Sorry for my usual longwinded blah, but here goes...

What I would most like to see for Puppy 6 is that it be platform independent as much as possible. ie: that it looks and behaves quite similarly whether you use it on a desktop PC, a laptop, a netbook or a tablet or on a Raspberry Pi.

This means that it needs a modular structure with a simple core, and a series of add-on modules that tailor it to the hardware it is running on. In a way this is the same sort of approach that Puppy has used with the “Wizard

User avatar
tronkel
Posts: 1116
Joined: Fri 30 Sep 2005, 11:27
Location: Vienna Austria
Contact:

#34 Post by tronkel »

@greengeek

Yes indeed, you could look at it like this:

Puppy = Core + A Collection of GUI Wizards
= Core + CGW as an abbreviation

There's no point in providing just a core and then saying to the user "go and install your own stuff" a la Arch Linux. Puppy needs to be fundamentally friendlier than that. There needs to be a friendly "welcome-screen" included along with the core, that points the user towards these wizard options .

This sort of welcome-screen approach has already been used in, amongst others Puppy Lucid, but I don't recall seeing it used on top of a bare bones core Puppy.

Also, as greengeek says, many of these "wizards" already exist anyway - they only need to be pulled together under a nice friendly front-end to make accessing them simple and obvious to inexperienced users who are doing an initial set-up of their Puppy system(s).

Good SW engineers, project leaders and designers think modular - so such use of wizards and sfs files would encourage this sort of approach right from the concept/design stage.

It's like starting an oil painting. The first brush stroke actually defines the final brush stroke. Make a mistake with the first stroke and you may never get to the final one without having to tear the thing up and then having to start from scratch. LOL it's like dejavu all over again!

The zdrv is simply another sfs module that contains extra drivers. The developer has the choice of building his system with this being either, an entirely separate module or, as an alternative, being included with the main pup.sfs module. These installable kernel modules would sometimes need to be recompiled to match a given kernel. If the core changes to a different kernel, then an updated zdrv.sfs would be needed with recompiled drivers.

This also leads to an interesting concept. I wonder if an updated kernel itself might be supplied in the form of an sfs file, to which the equivalent zdrv.sfs would also need to correspond.
Life is too short to spend it in front of a computer

User avatar
jemimah
Posts: 4307
Joined: Wed 26 Aug 2009, 19:56
Location: Tampa, FL
Contact:

#35 Post by jemimah »

Saluki is pretty much there now.

Core and apps are in separate sfses.

The iso is not a barebones ISO, thus the user can see the potential. However, you can get a barebones iso easily by removing the applications sfs.

The PPM is enhanced and polished.

Saluki custom builder is not a remastering tool - it's more like Unleashed on steroids. You can rebuild the applications sfs with only applications you want. You can also install the kernel of your choice and even cut it down to only drivers you need. Dependencies are checked and automatically downloaded.

Printing is modular. Either install the all-in-one sfs or install just the drivers you need from pets.

einar
Posts: 161
Joined: Fri 12 Nov 2010, 12:22

#36 Post by einar »

jemimah wrote:Saluki is pretty much there now.

Core and apps are in separate sfses.

The iso is not a barebones ISO, thus the user can see the potential. However, you can get a barebones iso easily by removing the applications sfs.

The PPM is enhanced and polished.

Saluki custom builder is not a remastering tool - it's more like Unleashed on steroids. You can rebuild the applications sfs with only applications you want. You can also install the kernel of your choice and even cut it down to only drivers you need. Dependencies are checked and automatically downloaded.

Printing is modular. Either install the all-in-one sfs or install just the drivers you need from pets.


in smaller words : its impressive

User avatar
Iguleder
Posts: 2026
Joined: Tue 11 Aug 2009, 09:36
Location: Israel, somewhere in the beautiful desert
Contact:

#37 Post by Iguleder »

Although I agree with you that Saluki does many things right, there's some annoying truth about it: it makes it even harder to build the first 64-bit or ARM Puppy and I'm intentionally ignoring FatDog and Lighthouse since they were not built using Woof.

With the huge amount of manually-built 32-bit packages Saluki has, it will be a nightmare to recompile all of them. I wish you good luck with that, if you want to port it.

I believe in good infrastructure. The right way to build a distribution which supports multiple architectures is having an architecture-independent infrastructure, which is something Puppy doesn't have - it explicitly loads many x86-specific drivers, relies on ancient x86 binaries and doesn't have any means of automated package building.

The right way to build a modular Puppy is starting from scratch, with a new, more efficient skeleton with less code, better documentation and better coding style. All those attempts to port Puppy the way it is are useless.

Time to confess - Puppy is getting old and fat, with big amounts of deprecated code and dependencies.
[url=http://dimakrasner.com/]My homepage[/url]
[url=https://github.com/dimkr]My GitHub profile[/url]

User avatar
jemimah
Posts: 4307
Joined: Wed 26 Aug 2009, 19:56
Location: Tampa, FL
Contact:

#38 Post by jemimah »

As stated above - saluki isn't trying to be puppy 6. I dislike monkeying with build systems. Porting isn't my thing either.

I have no intention of investing the time to do things "right". In 5 or 10 years, I'm sure puppy 6 will be awesome. ;) In the meantime, saluki is here, today.

User avatar
Lobster
Official Crustacean
Posts: 15522
Joined: Wed 04 May 2005, 06:06
Location: Paradox Realm
Contact:

#39 Post by Lobster »

Long term 'fatties', Racy and Wary and innovative Saluki are near completion :wink:
http://puppylinux.org/wikka/PuppyVersionIndex

Whilst I am still sitting up and begging for ARM powered Pi
http://puppylinux.org/wikka/PARM for future potential . . .
Puppy 7 also needs considering . . . :shock:

Cheap touch screen enabled devices be a coming . . .
bigger phones, smaller tablets . . .
and some major innovations in e-ink, flexible screens,
cheap Dragon (MIPS) chips, battery improvement
and flying penguins . . . and that is just this year . . .
http://echlinm.wordpress.com/2012/03/23 ... nd-drones/

Voice recognition is now advanced enough for
'no training' search engine input' at the least
but is Android app store integration even being considered?
Puppy Raspup 8.2Final 8)
Puppy Links Page http://www.smokey01.com/bruceb/puppy.html :D

User avatar
tronkel
Posts: 1116
Joined: Fri 30 Sep 2005, 11:27
Location: Vienna Austria
Contact:

#40 Post by tronkel »

I doubt if "right" could ever be achieved in the distro-building world. Even if both Jemimah and Iguleder came up with perfectly "right" concepts, their time of existence in the "right zone" would be finite i.e. limited to a certain time period. Distros are always evolutionary in their development and always will need to be re-vamped on an on-going basis - no matter how "right" they appear to be at any given point in time.

If you accept this view, then "right" is only a relative term.

There is no distro in existence at present which might be thought of as "right" - Puppy included. I think you can only talk about Puppy being sufficiently "right" for it being able to tank the opposition - even the big Linuces. It's the result that counts - how you got there is actually not really relevant - to the end user at least.

So Jemimah's stuff is sufficiently "right" for all practical purposes as of now, but to ensure that Puppy versions remain fairly and squarely in the "right" zone in the future, contributors and developers such Iguleder will be crucial to Puppy's ability to remain in that zone. Dunno though, maybe it will get to the point where at some stage the best way forward would be to start all over again with a blank sheet of paper.

Look at the classic case of MS Windows. It's now so slow that it's barely usable. The whole thing needs to be rebuilt from scratch for it to have any chance of staying level (technically speaking) with Linux. Think of how "right" Windows once was in say, 1995 or even earlier? Compare that to where it is now. LOL I think they might need Iguleder to go and sort it out for them!
Life is too short to spend it in front of a computer

User avatar
Iguleder
Posts: 2026
Joined: Tue 11 Aug 2009, 09:36
Location: Israel, somewhere in the beautiful desert
Contact:

#41 Post by Iguleder »

jemimah wrote:As stated above - saluki isn't trying to be puppy 6.
Fair enough.
jemimah wrote:I dislike monkeying with build systems. Porting isn't my thing either.
That's a bad design decision. In 5 or 10 years, you'll lose precious time with your family and other truly important things, because you'll be stuck in an infinite, manual compilation loop. Regarding porting - you can spend an entire lifetime trying to port Saluki - say goodbye to ARM and x86_64.
jemimah wrote:I have no intention of investing the time to do things "right".
Which means you're more than willing to waste your time on tasks that can be automated easily. What a waste!

jemimah wrote:In 5 or 10 years, I'm sure puppy 6 will be awesome. ;) In the meantime, saluki is here, today.
I can say the same - but with the "doing things right" thing - I built roar-ng for this. And it works great, today, tomorrow and it should work (e.g build packages, distributions and even assist in porting) even in 5 or 10 years from now, because of decisions I took early in its development.
[url=http://dimakrasner.com/]My homepage[/url]
[url=https://github.com/dimkr]My GitHub profile[/url]

User avatar
Aitch
Posts: 6518
Joined: Wed 04 Apr 2007, 15:57
Location: Chatham, Kent, UK

#42 Post by Aitch »

Iguleder

Leaving differences of opinion aside....may I ask if you'll be porting to arm at all - using the tools you've built?

I am sorry, but I get complete brainfade at any attempt to understand/absorb or act on your work, though I can sense the essence of it - which is very frustrating, having read and re-read several times....something just doesn't click in my brain.....it makes me feel like nooby, so I leave it to guys like you

thanks

Aitch :)

User avatar
Iguleder
Posts: 2026
Joined: Tue 11 Aug 2009, 09:36
Location: Israel, somewhere in the beautiful desert
Contact:

#43 Post by Iguleder »

In the long-term, maybe I'll make an ARM port, yes. I just don't have the hardware for this :cry:

I already ported my distro once, from x86 to x86_64, to test roar-ng and the platform it provides. I had to edit a configuration file (replace "i486" with "x86_64") and wait about 6 hours for the automated compilation of all my packages - that's it.

I'm 100% confident there won't be any Woof-based ARM or x86_64 Puppy anytime soon.
[url=http://dimakrasner.com/]My homepage[/url]
[url=https://github.com/dimkr]My GitHub profile[/url]

User avatar
jemimah
Posts: 4307
Joined: Wed 26 Aug 2009, 19:56
Location: Tampa, FL
Contact:

#44 Post by jemimah »

Iguleder wrote:
jemimah wrote:As stated above - saluki isn't trying to be puppy 6.
Fair enough.
jemimah wrote:I dislike monkeying with build systems. Porting isn't my thing either.
That's a bad design decision. In 5 or 10 years, you'll lose precious time with your family and other truly important things, because you'll be stuck in an infinite, manual compilation loop. Regarding porting - you can spend an entire lifetime trying to port Saluki - say goodbye to ARM and x86_64.
jemimah wrote:I have no intention of investing the time to do things "right".
Which means you're more than willing to waste your time on tasks that can be automated easily. What a waste!

jemimah wrote:In 5 or 10 years, I'm sure puppy 6 will be awesome. ;) In the meantime, saluki is here, today.
I can say the same - but with the "doing things right" thing - I built roar-ng for this. And it works great, today, tomorrow and it should work (e.g build packages, distributions and even assist in porting) even in 5 or 10 years from now, because of decisions I took early in its development.
My investment isn't in Saluki. I will abandon it when I get bored with it. The reason I do Linux development is to keep my brain active because I'm being paid to not work at the moment - and that should be a good thing, not a death sentence. The investment is in the wetware, not the software.

I'm a generalist, not a specialist. My superpower is being able to debug anything and get a lot done in a short time. The flip side of that is that I have a short attention span. I need to work on short projects under tight deadlines or I get bored and lose my momentum.

When I feel like starting over, I'll adapt whatever build system I think is best and have a distro in a few months. Not of the previous time spent is wasted, because it's all still in my head.

Some people are architects and some are field engineers - I am the latter. The world needs both.

User avatar
8-bit
Posts: 3406
Joined: Wed 04 Apr 2007, 03:37
Location: Oregon

#45 Post by 8-bit »

The comment on roaring almost sounded like Iguleder is upset that his builder was not used to make Saluki and in my personal opinion, Jemimah had did a great job of coordinating the making of it.

I have tried roaring several times, following instructions, to try to build the default distro it is set to make and failed royally all times to make a resulting ISO that contained the applications.
Call me dumb, but until the average guy can use it, it needs more time in the oven as I do not consider it done yet.
I know other have had success with it and rave about roaring, but whatever I am doing wrong, I am continuing to do even with printing out the build instructions and following them to the letter.

If I more or less hijacked this post, I am truly sorry and my post can be moved or deleted. But I just had to get me feelings on roaring out and my great liking of Saluki!
A person converting from windows would feel at home in it.
One thing that I would like to see though is the bootup process hidden behind a Saluki screen to finish giving it polish.

User avatar
mavrothal
Posts: 3096
Joined: Mon 24 Aug 2009, 18:23

#46 Post by mavrothal »

Some people are architects and some are field engineers. The world needs both
Puppy needs both!

But what if the field engineers are actually designing without a blueprint, and the architects are designing as a goal and not as means to a high quality "product"?
What about if the two do not communicate or at least have common end goals?

The outcome will be (at best) an "one time good thing" that will stay within "a very little pond".
More often than not, not even that.
== [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] ==

User avatar
greengeek
Posts: 5789
Joined: Tue 20 Jul 2010, 09:34
Location: Republic of Novo Zelande

#47 Post by greengeek »

mavrothal wrote:Puppy needs both!
What i love about this forum is that all points of view have an opportunity to contribute and shape new directions.

The forum is like a backbone - it provides a strong central core but enabling offshoots. There are many offshoots of puppy and only time will tell which of those has the best overall structure and longevity. Or even which offshoots will not even be puppies at all. There is room for everything.

At least the debates here offer a kind of "trial by fire" that subjects each version to wideranging scrutiny.

I'm just thankful that the devs snd troubleshooters have vision and energy.

ICPUG
Posts: 1308
Joined: Mon 25 Jul 2005, 00:09
Location: UK

#48 Post by ICPUG »

Iguleder is destined to be disappointed with Puppy!

The Puppy development method does not allow for infrastructure and gradual progression from one version/port to the next.

The Puppy development method is to:

- find a developer that wants to do a specific version.

- the developer then produces something while having as much fun as possible

Any suggestion about putting order into the development process is usually shouted down as not being fun (which is probably true).

I admire Iguleder's confidence in woof being the one true way and not looking at non-woof based puppies. He misses out on a lot of good ideas in my opinion.

If you want infrastructure then you need to be with a distro like Debian, Red Hat or SUSE.

Puppy development is chaotic. When a developer decides to build an ARM based Puppy it will be built. It probably will have tenuous links to any x86 Puppy but it will happen.

User avatar
jemimah
Posts: 4307
Joined: Wed 26 Aug 2009, 19:56
Location: Tampa, FL
Contact:

#49 Post by jemimah »

mavrothal wrote:
Some people are architects and some are field engineers. The world needs both
Puppy needs both!

But what if the field engineers are actually designing without a blueprint, and the architects are designing as a goal and not as means to a high quality "product"?
What about if the two do not communicate or at least have common end goals?

The outcome will be (at best) an "one time good thing" that will stay within "a very little pond".
More often than not, not even that.
Yes it's absolutely true. There's nothing for it but to wait until the right people are at the same place at the same time.

Post Reply