Whats up in Puppy World

[wiak]

Personally, I find MS Windows users unfriendly as an OS. So much crap in it and constant pop up warnings or messages, updates , or other major painful interference. I suppose some of that junk can be turned off, but all the long-winded wizards and so on are an admin nightmare to me.

Certainly, most of us, dont put a great deal of effort into 'polish' though some desktop environments such as XFCE or KDE start off more user friendly than I could myself hope to achieve starting with bare bones JWM or, say, Openbox. Puppy devs did marvels with JWM and fredx181 with Openbox designs, but such effort takes a long time and is huge.

[geo_c]

I have noticed several issues that no one pays any attention to or cares about. IMO, the real reason why Linux will never be mainstream is not due to the Microsoft monopoly, but to the general lack of user-friendliness in the OS/software. I think many, if not most computer users, would jump at the chance of leaving Microsoft, saving money, and ditching the surveillance/telemetry.

The misnomer here is that "Linux" is not user-friendly, as if Linux is a desktop OS brand, which it is not, it's basically a command-line kernel filesystem structure, Each individual linux distribution is an OS. So for instance a polished mainstream Linux distibution with a current KDE desktop, large maintained repo, and rolling release is every bit as user friendly as windows or mac, if not more. For comparison it would be better to say that Fedora w/KDE-plasma compares to Windows in a particular way.

As is demonstrated by this and other similar forum discussions, the puppy community is in a similar position. In other words, one would be more precise to say that vanillaDpup10 compares to Windows10 in a particular way, or KLV-Plasma compares to Fedora as far as its user-friendliness.

I'm saying Linux/Puppy is not an OS, it's a very large collection of OS's using a similar kernel and filesystem structures, and a plethora of common applications and libraries. It's the variety and fragmented nature of the Linux world that is confusing the issue about user-friendliness.

Personally, I find MS Windows users unfriendly as an OS. So much crap in it and constant pop up warnings or messages, updates , or other major painful interference. I suppose some of that junk can be turned off, but all the long-winded wizards and so on are an admin nightmare to me.

Yes, I agree. I just came into several computers with Windows 10 being thrown out by a company. I used a small linux command line tool to reset the admin password and booted one of them up. Windows is a pain in the ass. You can't really customize it's look convincingly, and navigation is painfully slow, because "user-friendly" means annoying nag screens and prompts to do this that or the other thing I have no interest in.

The "linux" customization abilities and user control is precisely one of the aspects that makes "user-friendliness" only attainable for large and well funded linux organizaions like Red Hat. It takes teams of people writing code to hold the users hand, and it's that same code which is nothing but an annoyance for computer users that want control over the machine.

And of course help dialogs and documentation only apply to a specific OS (or application), so as is the case with our forum distros, the variety and number of OS's makes the notion of complete user-friendliness for the largely computer or linux illiterate population an unlikely attainable goal.

And the addition of that sort of user-friendliness would likely motivate me to drop an OS for another.

Certainly, most of us, dont put a great deal of effort into 'polish' though some desktop environments such as XFCE or KDE start off more user friendly than I could myself hope to achieve starting with bare bones JWM or, say, Openbox. Puppy devs did marvels with JWM and fredx181 with Openbox designs, but such effort takes a long time and is huge.

KLV-plasma using KDE desktop is very intuitive to use. But it's not a tutorial in how to configure a Linux system for "atypical" uses.

Those of us that want the freedom to configure atypical setups are responsible for understanding a certain amount of what's going on under the hood.

I don't see that changing, and I don't think I want it to. If I wanted a monolithic, universally consistent OS environment, then I would use one of the "big guys" offerings.

[BologneChe]

@geo_c

What do you mean by "user-friendliness"? compared to other types of operating systems?

[geo_c]

@geo_c

What do you mean by "user-friendliness"? compared to other types of operating systems?

In response to @Governor and in light of what I have seen him struggle with, I mean by user-friendly that the OS behaves in a consistent and predictable way, and that help files and dialogs warn of every possible pitfall before performing a specific actions, and that compatibilty issues are overtly prohibited.

For example, If using puppy-linux A, built with slackware packages, and a user wants to install a debian package, does the OS explain in no uncertain terms to the user that these package are compiled differently? Does it warn, prohibit, explain how and when to compile?

Or if one wants to run an appimage on an OS only using Open Box as a window manager, Does the OS explain that one needs a gtk based desktop environement for it to run?

In other words, user-friendliness in the way I'm using it, and the way I'm assuming @Governor is using it, means that a person can run their operating system without having to dig for information and learn by trial and error.

That sort of thing is possible, if the user chooses a Linux distribution with a complete installer that reformats a drive and installs the OS to a single dedicated partition (or entire drive), completes the process from start to finish, doesn't give you the frugal install option, doesn't allow you to install anything that would break crucial libraries, etc...

[Governor]

@geo_c

What do you mean by "user-friendliness"? compared to other types of operating systems?

If you know, you know. If you don't know, you don't know. It is what it is.

[BologneChe]

@geo_c

What do you mean by "user-friendliness"? compared to other types of operating systems?

If you know, you know. If you don't know, you don't know. It is what it is.

A man of great wisdom...
I'm bleeding from my nose.

[geo_c]

@geo_c

What do you mean by "user-friendliness"? compared to other types of operating systems?

If you know, you know. If you don't know, you don't know. It is what it is.

A man of great wisdom...

Let me put it another way.

Is it possible to quadruple boot Windows 11, Mac Sonoma, ChromeOS, and EasyOS on a recent ChromeBook device? or a recent intel PC with Win11 pre-installed?

Maybe, but do you expect any of those OS's to include explicit instructions and installer routines to acheive it?

Only an experienced power user would try to do that sort of thing. And they would have to RTFM a lot.

That's one aspect of user-friendliness that Mac and Windows take care by making it as much of an impossibility as they can.

User-friendly in this sense, means predicting every possible user case scenario and writing help, dialog, and preventative measures to keep the user from borking his OS.

I'm saying it's not a desirable goal, and it's a waste of resources for a single or couple of puppy developers to pursue when they should be working on the under the hood stuff that makes the OS powerful.

[rockedge]

It's OK to have side gigs, but everything here looks like side gigs to me, while @wanderer is asking who's doing the main show.

Maybe because the "main gig" is to earn money. The stuff modern economies demand to survive. So what my wife insists is a frivolous hobby wasting my time for making coins, is somewhat unfocused in overall goals. Goals I am heavily required to reach are set by those who employ me and pay me.

I am a guy who knows how to light sets for cinematography and stage performances. Not really a school trained programmer or operating system designer/engineer, so I can't help really at the levels we are talking about here.

I can do some basic OS design and am able to do actually quite a bit with programming and "such" but we need innovation and serious Dev's who know how.

I keep saying what the main dev path should be. That is the one OS that @dimkr doesn't want for it to be.

Even at Zoneminder the dev(s) now mostly do any cool stuff only for those who pay directly or donate directly. Not enough dev's and too much to do for just the "satisfaction" of a job well done.

[dimkr]

we need innovation and serious Dev's who know how.

I agree with that but I also think we need focus. Without focus, it's just wasted effort and stagnation. Look at that 4.3.1 video, Puppy hasn't changed much since 2009: one release adds conky and quickpet, the next one removes them, then this release switches to ptheme, then that one switches to jwmdesk ... it's mostly a zero sum game. (IMO Bluetooth support and apt are among the few 'killer features' of recent Puppy releases.)

Areas that do need improvement, like installation and package management, don't get all the attention they can get. Maybe because the few developers we have find them too messy and risky to change or not fun to work on (unlike new JWM themes and wallpapers).

[wanderer]

hi dimkr

it seems you are doing everything that you are suggesting

you are making a cutting edge puppy (or puppy-like ) distro

people can just get behind your work if they want to go in that direction

wanderer

[wanderer]

hi sonny

i agree with you

as to being completely unique
that is a delusion
everything comes from something else

all linux distros use the linux kernel

as to parts of puppy that need improvement
one can build what is needed from "scratch"
( but of course you will have to use a lot of what others have already made to do that )
or adopt it ready made from another distro
that is what linux is all about

i see the diversity in puppy as a good thing

there is a lot of development going on
puppy has changed a lot over the years and will continue to change

where puppy goes in the future
well thats the point of this thread

and there clearly is a lot of interest
we have 3549 registered members
and 274 hits per day on distrowatch

wanderer

[wanderer]

hi all

as many of you are painfully aware
i feel puppy should be made into a modular system

i have been looking around to see if i could find any distros that fit the bill
and have revisited slax which has improved since i last tried it

i am playing with slax debian 64
very interesting system

i encourage everyone to take a look at slax
since i feel that is the direction puppy should go

thoughts anyone

wanderer

[greengeek]

hi all

i would like to see a modular puppy

1. bootloader module

2. kernel

3. core (command line only) module

4. basic x module

5. sfs modules ( extensions for other applications )

i think it would be a useful base for further development

Could you give a little more information about how this would work and which part of this is lacking in puppy?
eg: What would a user do with a bootloader module? Would this contain all the necessary firmware to boot a PC as far as a command prompt? ie just waiting for step 2 the kernel load?

And would that step 2 kernel load be aotumatic or under manual control of the user?

What would the step 3 core module look like? Presumably just a command prompt? Or something more?

Step 4 is self explanatory with X loading - but would X have any menus or programs available? Or would that be step 5

Step 5 - would you include all puppy utilities and programs in sfs form? Or just the larger ones?

(I have some thoughts about this after having used Ozsouth's cut down Fossa 64 for a while - and I am trying to compare your plans with what PeeBee and Ozsouth have already made available...)

[wanderer]

hi greengeek

take a look at slax debian 64

this is exactly what i am talking about

we could tweak it / fork it to do whatever else we want

but it is finished as it is

i am exploring / playing with it now

and will talk more about it on this thread

wanderer

[fredx181]

as many of you are painfully aware
i feel puppy should be made into a modular system

i have been looking around to see if i could find any distros that fit the bill
and have revisited slax which has improved since i last tried it

i am playing with slax debian 64

Puppy is already a modular system (as all other OS's are too presented in this forum, btw)
Slax does have a more specified module setup though:

2024-08-04_10-27-55.png (19.29 KiB) Viewed 3986 times

Where in Puppy the modules are named adrv bdrv fdrv zdrv (and the main sfs)

But I don't really see a major difference with how Slax is setup, (and what can be the advantage of the Slax method) .

[wanderer]

hi fredx181

the key is --- how --- it is organized

every distro has a small number of basic components

bootloader
kernel
ramdrive

shell and utilities
package manager

video
sound
internet

window manager
terminal
editor
file manager

browser
media player
image viewer

other utilities

other applications

if these are broken into modules
they can be maintained individually and exchanged as desired

this provides major advantages for both the user and the developer

for the user
they can choose what works for their system and suits their tastes
and swap out what they want and don't want

for the developer
they can work on only the module that is needed/desired
without having to redo the entire system each time

a small repository of various modules can be maintained
so that assembling a distro is easy and simple

( for example by using your 1 script build system )

so someone could build a distro for an older computer
and someone could build a system for a newer computer
and customize it to their tastes
with most of the same modules being reused for each

wanderer

[wiak]

Of course, a distro like that could be assembled here, and yes there are obvious attractions to that approach.

The reason why most designs do not use that apparently simple plugin style modularity is because it comes with a major flaw in terms of practicality: inherently it breaks reliable package management. By reliable I am talking about package management systems that are expected to accurately maintain database of the whole chain of dependencies involved in any package installation. In typical accurate package management it is impossible to take a bunch of packages out, replace them with a completely different set that were assembled into a module perhaps months previous. Such swapping in of modules breaks the package manager database such that the system no longer knows what is has installed nor if new libs and so on remain compatible. If we could do it we would have done it long ago (well it was done - traditional Puppy being an example of anything goes almost 'package managent').

Well, so how come some little distros do provide such a modular arrangement? Simple answer is that they either use only their own specially built/compiled repos that always ensure diffrerenr modules remain compatible with each other, or, more like teaditional Puppy they do not bother with near perfect reliable package management at all, which makes life simple and fun but often results in great difficulty getting all new app installs to even work like alone work perfectly without error or system freezes and so on.

Puppy users tended to enjoy random assemblies of pet packages, hand crafted sfs almost portable apps, and alien pacages in .deb or .rpm or .tar.gz form and so on. But the problems and unreliability of, for example Puppy PPM resulted in constant issues and rough workarounds and broken systems such that some forum members desired replacement of that chaotic modular anything goes arrangement and wanted powerful reliable dpkg/apt, or say, Void Linux xbps, instead. But then you get the broken package databsse result if you install anything, such as an sfs module, that doesnt get provided by the upstream package manager (e.g. dpkj/apt). The more reliable upstream package management al also ensures correct security updates.

So which approach is really better?

In short, yes can make modularised system, but has huge drawbacks touched upon above.

[wanderer]

hi wiak

yes i understand

im talking about a very very limited number of modules ( less than 20 )
most of which have already been made
and that will not have to change much at all over time
based on only 1 system ( debian 12 )
with apt as the package manager
that can serve as a base distro

however most people will never need anything more than this base for their day to day use

people can add whatever else they want later
but its up to them to make it work

i always felt it was very ambitious that puppy tried to make systems out of every distro under the sun
and its amazing that the gurus were able to do it most of the time
but that never made any sense to me
you will never be able to fine tune a system like that

it also important to be aware that i am not asking anyone to do this
this is just a suggestion that i think might benefit the puppy community
and be useful and interesting

in fact it has already been done
so no one needs to do anything

slax does this very well
and tinycore does it
(though not so well but only because the extensions are not always user friendly )

but since this thread is about ideas that might improve puppy
there is is

wanderer

[dimkr]

if these are broken into modules
they can be maintained individually and exchanged as desired

this provides major advantages for both the user and the developer

IMO maintaining separate "modules" means way more work for developers. Debian isn't "modular" and it will take a lot of work to divide it into "modules". If it's a one-time manual effort, then you're stuck with something you can't update easily and you'll need to repeat this work if you want to update (and you want to update if you care about security and stability).

And there are disadvantages - each "module" consumes extra memory, and you'll achieve a much worse compression ratio if you split your OS into many "modules" that are compressed individually.

for the user
they can choose what works for their system and suits their tastes
and swap out what they want and don't want

Users can swap kernels even today.

for the developer
they can work on only the module that is needed/desired
without having to redo the entire system each time

Building a distro based on Debian 12 and leaving it mostly unmaintained until Debian 13 is out won't save time and make development easier. It will take you a very long time to rebuild your distro against Debian 13 if you don't do any development for 2 years after you made your Debian 12 based release. For example, if X.Org is gone in Debian 13, you'll need to port everything to Wayland before you have a complete, working distro, and this can take a very long time. Continuous development in small chunks of work is easier and produces better results because you have time to polish things.

a small repository of various modules can be maintained
so that assembling a distro is easy and simple

This goal can be achieved by writing a good build system that is easy to use, this doesn't require "modules".

so someone could build a distro for an older computer
and someone could build a system for a newer computer

How do you do that when they all share the same base? Debian 12 supports some range of GPUs and you'll need to fork Debian if you want to replace the kernel, Mesa, etc' to build a "Debian for old computers" or a "Debian for new computers". Reality doesn't consist of LEGO parts you can swap freely, this so-called modularity is impractical.

[wiak]

In my view, that modular idea simply doesn't really work in practice. I've thought about it many times, but always rejected it in practice as more trouble and time-consuming than actually worth. The only exception would be a small set of modules in their own kind of repo, but no other package manager or at least no dependency resolving package manager with a database - but the idea then is simply to not attempt to every upgrade these modules or only 'eventually and all together...'; overall the approach is not attractive at all except as a phony marketing play that has no read advantage or meaning.

So could make a distro comprised of 20 modules and it would work as long as you never want to upgrade it... In theory, yes same distro could allow for some such modules to be removed or a few other modules added, but none of that upgrade-able or the whole thing basically falls apart or involves far too much work synchronising upgrade of all parts, or have a really quite unreliable package manager (like PPM) that allows suck it and see if still works... and often it will and often it will seem to, but inherently be broken, which user will painfully find out now and then. I hesitant to say - is that not a bit like traditional Puppy?, but made more 'modular' and even harder to keep from becoming broken. Yes, keeping such modules 'upgraded' and able to work together is a lot of time-wasting dev work - not easier in the least at the end of the day. Sorry, wanderer - I see that as a utopian dream that is nonsense in reality, whereas non-modular (in that sense) but with really reliable dependecy-resolving package manager does in practice work like a dream and is much much easier to maintain.

It is simply not a bad design 'accident' that most practical good useful distros chose an overall reliable package manager approach and not a piece by piece modular glue-able part approach unless you go for big Flatpak, Snap, or AppImage approach, which does not result in any small to download type of distro at all (but can be made to work really well and with probably best reliability of all, if you don't mind big download size).

[wiak]

A build system that is modular is a different matter altogether, and some distro items are able to be used in modular fashion such as alternative initrd, alternative kernel with suitably matched modules and firmware, and modular overall root filesystems - different offering for different desktop environment, just for example. But NOT modular root filesystem - aside from occasional portable app type arrangement and, as I said, bigger systems involving FlatPak, Snap or AppImages. At the end of the day a package manager that takes into account the whole root filesystem is key, but of course you may want to freeze such items as the kernel/modules/firmware such that they do not get upgraded else employ a flexible ability to upgrade the kernel/modules/firmware, including any necessary re-building of the initrd, at the same time such upgrade is performed on the overall root filesystem.

[wanderer]

hi guys

thanks for taking the time to explain all this
i now am able to see the pros and cons much more clearly
i think i will give up on the modular stuff

but that leaves me with a question

in your opinion what is the best overall system to build a distro

wanderer

[wiak]

in your opinion what is the best overall system to build a distro

Sorry wanderer, I really don't understand that question. All I could likely answer is that 'best' is always likely to (indeed) simply be a matter of opinion and even such opinion depends on what you are trying to do or achieve, and will still be a matter of opinion probably. But actually I really don't understand what you mean.

[wanderer]

i know opinions and cases differ
but lets say what you would do to build a puppy-like distro

i like fredx181s single script to put all the pieces together

but generally it seems like the key is the package manager

wanderer

[fredx181]

i know opinions and cases differ
but lets say what you would do to build a puppy-like distro

i like fredx181s single script to put all the pieces together

but generally it seems like the key is the package manager

It's not very clear to me too what's your question, is it still about a modular system as you described earlier (e.g. with 20 modules) ?.
Basically it should be possible to make a Debiandog alike system that way, but I agree with wiak that it has disadvantages and that many problems can appear, e.g. breaking the package management (which would be disastrous).
Just one main module (and kernel module and perhaps firmware module ) is the best IMO, as the package manager then "knows" exactly about what packages (and which versions) are inside.
edit: but you can of course add more modules in debiandog , but then without package registration (to prevent breaking apt/dpkg) (can be created that way with "apt2sfs").

[wanderer]

thanks fredx181

thats the answer i was looking for

"Just one main module (and kernel module and perhaps firmware module ) is the best IMO,
as the package manager then "knows" exactly about what packages (and which versions) are inside."

wanderer

[fredx181]

thanks fredx181

thats the answer i was looking for

"Just one main module (and kernel module and perhaps firmware module ) is the best IMO,
as the package manager then "knows" exactly about what packages (and which versions) are inside."

wanderer

To get back about Slax, question could be "How come that Slax works ok with it's 5 separate modules ?"
Well, my guess is that an average user can break Slax's package management easily when modifying that modular setup, by exchanging modules and/or add extra modules.
(so, in other words, the multi-module setup looks nicer than it is !)
Fine of course if you use it as it's prepared, and use save file or folder for the changes.

edit: with very early versions of Slax (based on slackware) it was different, but nowadays Slax is based on Debian and dpkg/apt is very strict.

[dimkr]

The question is not "what can I do with this distro" but "what can I do with this distro that is not a terrible idea". Playing with "modules" is fun until you reach a state where you can't install anything anymore and must go back to the initial state, or until the distro can't be developed further because it was built in an error-prone, laborious, manual process that emphasizes modularity over practicality and future-proofness of the project.

[wanderer]

hi all

anyone who has been reading the forum and following the puppy community
can see that the puppy community has never been more active and innovative
puppy has become a virtual distrowatch of innovation

and i have been around a long time
and know what im talking about

unfortunately there are negative people who spout nonsense
normally i would ignore them
but in this case i will make an exception
they need to get a dog to kick ( hopefully not a puppy )
and learn to post on the right thread

let us all thank rockedge (who has made a herculean effort in time work and money
and all the puppy community ( who i will not try to name because i might leave some out )
both developers and users
for all that they have done
to not only keep puppy alive but to make it thrive

wanderer

[Wiz57]

thanks fredx181

thats the answer i was looking for

"Just one main module (and kernel module and perhaps firmware module ) is the best IMO,
as the package manager then "knows" exactly about what packages (and which versions) are inside."

wanderer

To get back about Slax, question could be "How come that Slax works ok with it's 5 separate modules ?"
Well, my guess is that an average user can break Slax's package management easily when modifying that modular setup, by exchanging modules and/or add extra modules.
(so, in other words, the multi-module setup looks nicer than it is !)
Fine of course if you use it as it's prepared, and use save file or folder for the changes.

edit: with very early versions of Slax (based on slackware) it was different, but nowadays Slax is based on Debian and dpkg/apt is very strict.

Just an FYI...you can also get Slax based on Slackware, AGAIN!!! In addition to the Debian version.
Slax, both variants as much as I am aware, uses an internal packaging system, you can create "sb"s
(Slax bundles) with tools included in the ISO. I used to do that many years ago, and the directory
structure of Slax allows you to include all your sbs at boot time, OR you can optionally load them
when wanted by placing them in /slax/opt directory (or is it named slax/optional...can't recall right
off the top of my head). Not sure if there is a limit to number of slax bundles you can load, probably
not, only to the limits of your RAM.
Wiz