Community Edition anyone interested?

[darry1966]

..As for Jpeps I think is absolutely correct in what he is saying about a modern Pup it should meet with what is required now so as part of the community he is rightly voicing that....

Yes please, start building something for modern computers.
When you will have finished it, this computer will also be an old one.

Exactly the progress being made is very slow hence my fustration and Jpeps, lots of eloquent speeches which mean well.

Confusion over direction etc especially Retro computers as well and what the bases for development will be.

[jpeps]

..As for Jpeps I think is absolutely correct in what he is saying about a modern Pup it should meet with what is required now so as part of the community he is rightly voicing that....

Yes please, start building something for modern computers.
When you will have finished it, this computer will also be an old one.

Frankly, I don't see the point of trying to "puppify" something like an android device that has access to a million apps and is specifically adapted to touch screens. Better tools may become integrated as available. Until then, what's wrong with the current method of just fixing and improving what is already available? Devices are now moving to larger screens with more serious applications, so will become increasingly competitive.

MS and Apple are competing with the shareware model by giving away software. Prices should continue to come down.

[mikeslr]

Hi All,

Having noticed that each new class of Pups is slower than the last (and I think wireless less capable of picking up weak/distant signals) , I've been reading thru the pUPnGo thread.

If I understand correctly, a big if, the most significant factor which adversely impacts the speed and size of Pups has been the increase in size of glibc and Gtk. Someone, I forget who, recently posted that Pups can use any Gtk version, indeed, several versions. And if I understand amigo's post, glibc is not kernel dependent but can be compiled under any kernel.

Goingnut's method in pUPnGo was to strip a 4.12 pup to a bootable core, but provide user-functionality by placing applications in zdrv which would be loaded at bootup. But he began his work before jemimah developed the adrv. [Jemimah was aware of goingnut's work, and may have been inspired by it]. After that development, Pups could be built to load at initial bootup both an zdrv (which traditionally only contains drivers) and an adrv which contained applications. Thereafter, additional applications could be added by the user via SFS load and/or pet install, (or in Saluki and Carolina using custom-builder to build a new adrv).

So I wonder if, rather than multiple Pups (other than one 32-bit non-pae and one 64-bit), the following would be possible:
(1) a “core

[slenkar]

I agree with mikeslr

[mikeslr]

Hi again,

It was suggested to me via email that "mix and match" could result in a mess. On reflection, my guess is that's probably true regarding glibc. At any rate, it's an unknown, something best tested other than in a Community Edition. On the other hand, we already have pups which use a couple of versions of Gtk. So I'd amend the previous post to provide only for Gtk SFSes for "mixing" with the warning that (a) just adding a higher number gtk may not be sufficient to support applications dependent on a different adrv and (b) SFSes can be unloaded if they generate problems. Users can also be advised that a possible consequence of changing adrv's is that applications installed to a SaveFile (or remaster) may break and application SFSes may not function if they required the glibc of a different adrv.

mikesLr

[darry1966]

Sounds like an interesting idea Mikesir however this is a community distro thread and the idea is to produce iso's 1. for retro whatever that is as long as it offers support for more than a few types of retro machines an updated 4.32 a distro that had a wide base of support and video capability that was ideal enough for old kit would be ideal and something to fit with modern needs so lets keep our heads and feet on the ground and not keep this thread going as some kind of wish list that unfortunately it has become as Jpeps was more or less saying between the lines "keep it realistic" - Get a good base for each going and then the testers having something to work with we ain't a concord.

[technosaurus]

Just make the devx.sfs have the oldest supported version of glibc, gtk, etc... End of problem. If a developer/packager wants/needs to build against a newer version, then they will have to go through minor hoops he/she should be prompted to mention it - preventing a lot of common issues.

[jpeps]

Just make the devx.sfs have the oldest supported version of glibc, gtk, etc... End of problem. If a developer/packager wants/needs to build against a newer version, then they will have to go through minor hoops

Minor hoops like it won't compile?

[technosaurus]

Just make the devx.sfs have the oldest supported version of glibc, gtk, etc... End of problem. If a developer/packager wants/needs to build against a newer version, then they will have to go through minor hoops

Minor hoops like it won't compile?

... _may_ not compile (many/most will) unless you update packages, but if you are going to make newbs update their packages, then you should at least be able to make them aware ... having to update it themselves would make packagers aware that it would not be supported in some versions without updates so they can pass along the info. ... for example abiword would compile with gtk+-2.12 (and worked better in Puppy due to omitted gvfs) but if compiled in gtk+-2.14 or greater it would not work on older versions at all and many functions were broken if used with gtk versions >= 2.14.

[gcmartin]

This post tries to understand a prior quote:

Having noticed that each new class of Pups is slower than the last (and I think wireless less capable of picking up weak/distant signals) , I've been reading thru the pUPnGo thread.

I have NOT noticed this. And, I question how this could be true. Each PUP even those which are based upon other's base, come to us with its own characteristic behavior. Much of that is derived from advance functionality via kernel while other is due to advanced ability in both subsystems and application (which include utilities). Puppy, IIRC, is fast....stable for the most part ... and highly functional and usable.

I find this an interesting comment as most every PUP ships with a utility which shows performance. It is easy to run. And, I am not so sure whether the quote refers to a particular application or the desktop or LAN services when read-write to LAN PCs or web/multimedia serving or ???

I have NOT observed this nor have I seen anything widespread in the community reporting that suggest such.

Lastly, if an application/subsystem has been enhanced is such a way as to increase functionality to the user with little penalty, this couldn't possibly mean that the application is bad.

Puppy runs very very well when matched against other industry Linux distros.

How many of this community have also seen a "degradation" in performance of PUPs?

If so, should the CE start with that as an item for new?

[gcmartin]

As I think about it, maybe its time to do one additional thing for common distro releases: produce some guidelines or document to address the traditional problems ALL Linux(s) have; namely

• Video issues
• Audio issues
• Wireless issues/LAN issues/Modem issues
• Keyboard-mouse issues

Question
Would each distro need its very own document/guide/threads or are the resolutions of such issues something which is reasonably standard as to be addressed similarly in ALL PUPs, including CE?

[mavrothal]

So I wonder if, rather than multiple Pups (other than one 32-bit non-pae and one 64-bit), the following would be possible:

This is certainly possible and not a bad idea either, however there are plenty of devils in the details that may make things difficult.
Currently mainstream puppies use 3-4 different toolchains. Puppy in that respect is a miniature of the GNU/Linux echosystem with its zillion of (incompatible) distros, only worse because the common name makes you think they are compatible.
The second is the binary compatibility. Puppy is not 100% compatible with any other distro, but T2. But then you have to build the 10000 packages the other distros have and still some is hard or impossible to build in T2.
Packages, either from 3rd source or T2 need serious quality assurance and dependency checks done either at the time of packaging or by the package manager. This is more important if you are to follow the multi-sfs system. Saluki was using only its own well audited (and limited number) packages to build the adrive. With 3+ SFSs this becomes a nightmare since you have to account for many possible combinations.
All these have to be automated at build time by a builder script(s) (woof or otherwise) that should take care of the above issues. Do it manually is much more difficult to maintain (I did that in the original XOpup, I know) and is gone with the developer

To do all these you will need a team of a minimum of 5-10 capable, dedicated and financially independent people to put the thousands of work-hours needed (most big distros have employed full-time developers at their helm)

If you do all these, puppy will not be puppy anymore. Will either be a big independent distro or a distro spin/fork...

But the major question is what is that bother you in the current puppy???

New vs Old computers: why should you have 1 puppy to fit in everything?
Different looks: why is a problem if say LXDE/XFCE/Mate etc is installed on top of jwm/rox? Most likely the computer has plenty of RAM and speed to go for these.
Package incompatibility: If you stay within the ppm offerings you are mostly safe. If you start installing whatever is around in the forum you run into problems. For real problems, the forum has been proven invaluable.

Is puppy perfect then? Far from it.
But I think is important building a CE puppy to focus on what you want to do with the build system. Big distros have what they call "use case" where you describe a real life scenario that is not satisfied by the current offering. If this is deemed reasonable then you try to find ways to solve it.
If for example someone says that I want to have one USB stick and with this be able to boot every computer from a PentiumIII to quantum computers( ) from console to GNOME4 ( ), use it also as a server and be under 200-300-400 MB, may not be feasible.
Specially if we are talking people that build puppy in their spare time, usually by themselves.

[mikeslr]

Hi all,

My apologies. I've said it before. Not every idea I have is a good idea. It was motivated by a desire to bring a little order into what has become, for a user, a chaotic mixture of distros operating under one name. Things may be fine if a user relies only on the applications provided by the dev, those in the repo created for the Pup, maybe those in repos suggested by the dev, or compiles his own. Install an app with a different lineage and you're asking for trouble. And as I also said, we are constantly re-inventing the wheel: building the same applications over and over again, albeit, for different "derivatives." And lastly, by constantly building and advertising "the new" we obscure and abandon the great work which was previously accomplished and still has a place for those seeking a operating system for "older" computers.
But as mavrothal said "The devil is in the details."
Beyond any compiling problem, attempting to actualize the idea produces two other problems: one logistic, the other informational. There are multiple ways of running Puppys. One way is to run it from a CD/DVD without a SaveFile. Not possible OOTB if the user wanted to change the adrv. The same applies to Full Installs. I also have no idea what effect changing an adrv might have on trying to run Puppy via Mult-session CD/DVD. The idea also creates further complexity in PPM, and repo management. Finally, and perhaps most importantly, while currently with all its faults, a Puppy can be easily put to use by a noob without any knowledge of Linux, or desire for it, 'Mix & match' would immediately present her/him with some complexity and require that we provide instructions and perhaps explanations. [I realized the above while attempting to write up in "laymen's terms" something to add to First Run dialog to explain the choices].
While the logistics and informational problems could be overcome --might, in fact, be worth doing-- they would initially create complexity rather than simplify,

gcmartin: I've run Puppies since 2.1x. With the exception of the computer sitting next to my router, all my computers access the internet via wireless. The router is on the second floor. My den/tv-room on the first, my main (most powerful: 4x AMD Phenom(tm) II X4 945 Processor, 4 Gb Ram), computer in the basement. I explore/test many Pups from my main computer as they are published. The Grub4dos menu.lst of my main computer lists raring, several precise variations, a couple of Slacko variations including FatDog64, and Lupu. On the first floor, when I set up wireless, the adapter sees 8 networks. Lupu, in the basement sees 3. The others see only my network. And I have to install “stay-connected

[wanderer]

just a question

why doesn't the community edition start with something like pupngo
(with goingnuts' approval of course)

1. we can start right away, because we will already have a base.
2. we can concentrate on one iso for the time being, so its more manageable.
3. the 4 series is good on old and new hardware.
4. its modular, so that we can add things slowly, as we get the time and expertise, for example, kernels, xorg, sfs files, program folders, etc.
5. we can woof it (one part at at time) as we go along, so it is saved and reproduceable.

starting from scratch and making a puppy that can do everything sounds great, but puppy already has all the systems needed to build a flexible distro

your thoughts

wanderer

[darry1966]

just a question

why doesn't the community edition start with something like pupngo
(with goingnuts' approval of course)

1. we can start right away, because we will already have a base.
2. we can concentrate on one iso for the time being, so its more manageable.
3. the 4 series is good on old and new hardware.
3. its modular, so that we can add things slowly, as we get the time and expertise, for example, kernels, xorg, sfs files, program folders, etc.
4. we can woof it (one part at at time) as we go along, so it is saved and reproduceable.

starting from scratch and making a puppy that can do everything sounds great, but puppy already has all the systems needed to build a flexible distro

your thoughts

wanderer

1. Yes if a decent Glibc, GTK can be added SDL support and possibility to add dri and GL support in Xorg like in normal series 4 PUPs for old machines. PUPNGO seems great for 486's and early pentiums.
2. Yes
3. More than one kernel available would be useful.

[gcmartin]

Please, let us start by stating, when release, what PCs we are aiming at. Pentium 1s hit the market around 1994...almost 20 years ago. Why would we want a 20 year old power hungry system as we march this Puppy into the future. Puppy Linux distros, long ago, addresses these. We have moved on and should NOT be looking at trying to make old PCs modern via software. It just cannot be done.

I fear that there are many members who have a wrong view of taking a low compute hardware platform and turning it into a modern beast capable of gobbling up many trillions of instructions each second. Puppy Linux (in fact NO Linux or any other operating system, which is software), cannot do this. Hardware either possesses this or not! The OS uses hardware to carry out the efforts we need for our needs. Our needs vary from text files, to internet to photo editing to video-audio production to you-name-it, Much of Puppy today addresses much of needs either in its base or via Package Management (PM). We still have people in the forum who find that pae is somehow wrong sighted when most EVERY PCs has it except for a very few during the time the industry was looking at netbooks. The 2 netbooks donated to me are both 64bit. Other than myself, NO OTHER MEMBER want to acknowledge that they have a netbook capable of these technologies...Wow? (excuse me, I believe Jemimah did, some time ago).

What do users look for when they come to Puppyland? I maintain they want a good base of applications and utilities where it is obvious in what and how to use the desktop. They profit from a simple understanding of what PC is expected for the PUP and where documentation exist to explain the PUP. And if they get 90% of their needs without having to address PM, then they are "happy campers" because of the nature of PUPPY's behavior running as a RAM based OS.

So far, it is my observation that no prior Windows users has come to this forum about any complaint in PUP performance when trying out their initial experience with Puppy. So, we need to build upon this fact and expand our successes in such a way as to make Puppyland an attractive kennel for them to want to be a part of. (In fact, I fear some will try to tear this paragraph apart while missing its theme.)

Let's be very clear about our distro, our documentation, and our direction.

And, lets dispel any fear in documenting what is done. In fact, let have all of us members here support the final direction taken and discount anyone who takes aim to discourage any direction taken. This does NOT say to be less than perfect; instead it means to be as stellar as we can possibly be and make plain what we are providing for community benefit.

The idea to be taken, at this point, is to take what comments already made, and appeal to those (especially the knowledgeable contributors in this thread) for their assistance with the techniques and offerings they have already suggested/provided.

And, let try to anticipate how to do this to make easy future expansion, future off-springs and future remasters.

Here to help

[tlchost]

What do users look for when they come to Puppyland?

What type of users? If exisiting Linux users...that's one set of wishes/requirements. If Windows users, and entirely different set. If exisiting Puppy users, yet another permutation.

If I were leading the charge, I'd concentrate my efforts on creating a Puppy that would make windows users very welcome....they are probably the largest class of computer users who are potential newcomers to Puppy.

If that's the case, then Puppy needs to run on modern equipment...as I suspect the bulk of Windows users are NOT using Pentiums or some other favored, revered and preserved technology.

Should we abandon old hardware? No. Should we cater only to Linux Newbees? No.

I think wwe have at least 3 directions to consider....Newbees coming from Windows using modern hardware, Users with older hardware and existing Puppy users.

Thom

[Rattlehead]

Hey guys, I've been following with great interest this conversation on Puppy's future. Here are my 2 cents:

*Allocate a solid chunk of time to define a vision before getting 'down to it'. I'd say for example 1 month from today. If the period turns to be too much time, thanks to Internet and the power of collective thinking, then so be it, but in my experience every time dedicated to this envisioning is worth every second, because it saves a lot of headache later and also because, once a steady and clear vision is established, there is no shortage of man-hours in the Linux community, and even more will join if they see a defined direction. The problem I see now is that there is a lot of good will people ready to do stuff, but everybody pulling in a different direction, and, as it is often mentioned here, perhaps reinventing the wheel in small circles.

*The vision should be defined in user-centric terms. It will help us focus on outcomes, and cut the technical discussion to the strictly necessary.

*My personal preference: something that has called my attention in these threads is that Windows XP will stop being supported by Microsoft in April 2014. It means half a billion(!) XP computers will start to 'fall to pieces' from that date, and it is easy to suppose that some users will be interested in finding a software replacement rather than dumping their machines. Why not addressing such need? It frames the computer generation to address, and perhaps opens a great opportunity to increase the Puppy user base.

Under such premises, my vision, in user-centric terms, would be something in the lines of:

"A comfortable replacement solution for 'homeless' XP users"

"A stress-less continuation of the basic functions offered by XP, free of its hindrances"

"A bridge distribution for XP users who want to get initiated in Linux".

Something like that. The best definition should be agreed and fleshed out before a single line of code is touched. The philosophy behind is 'going slow so that later we can go really fast'.

As I said, this vision is just my personal, probably uninformed preference, just a way to get the ball rolling. You are invited to say such vision is not worth to pursuit, as long as, constructively, you provide your modified version of it, or a different vision altogether.

I know perhaps this vision thing sounds like a 'chore', but I think it would save a lot of organizational hassle later. Just saying. Let me know your thoughts.

[wanderer]

i like the xp replacement idea. one could make a iso that runs on both (most) old computers and xp machines. i use a xp machine for 214x9 and it runs great.

by the way, i also run 214x9 on a windows 7 machine and it runs great

you could even advertise CE on distrowatch as an xp (and 7) replacement
that should get a lot of people involved in puppy

wanderer

[greengeek]

Windows XP will stop being supported by Microsoft in April 2014. It means half a billion(!) XP computers will start to 'fall to pieces' from that date, and it is easy to suppose that some users will be interested in finding a software replacement rather than dumping their machines.

A very excellent point. Perhaps this turning point for XP suggests that the CE project should include a really well functioning version of Wine - it is not just the OS itself that people will be needing to replace - they will want their programs to run well too.