Do we really need a NEXT PUPPY ?

This is where threads concerning the development of the next version of Puppy live.
Message
Author
Robin2
Posts: 180
Joined: Sat 17 Jan 2015, 18:17

Do we really need a NEXT PUPPY ?

#1 Post by Robin2 »

OK, maybe the title is a little provocative ...

But a big part of the reason I switched from Xubuntu to Puppy was to get off the "update treadmill"

I reckon there are folks building parts for Ubuntu (and everything else including cars and clothes) who say "that's finished - what'll we do next. OH, lets make another new version or else we'll have nothing to do"

Of course there is a need, from time to time, to bring out a new version when an error is discovered.

But, apart from that, I don't think we should be frogmarched into new versions.

Let the developers of new versions advertise the improvements and then I can decide if they are of any value to me. But don't include them automatically.

...R

User avatar
rcrsn51
Posts: 13096
Joined: Tue 05 Sep 2006, 13:50
Location: Stratford, Ontario

Re: Do we really need a NEXT PUPPY ?

#2 Post by rcrsn51 »

Robin2 wrote:But, apart from that, I don't think we should be frogmarched into new versions.
That's an odd phrase. No one is ever forced into a new version. Your old version will not spontaneously self-destruct when the new one appears.

But there are several good reasons for building new Puppies. New kernels mean support for new hardware. If you buy a new wifi adapter and try to run it in Puppy 4, chances are slim that it will work OOTB.

Third-party software developers often build their projects on the latest mainstream Linux platform. In order that those apps also run in Puppy, Puppy has to keep up with the mainstream.

If all your stuff runs fine in Puppy X, that's great. You have no need to upgrade. But that's not true for everyone. If some community members had their way, we would still be stuck on Puppy 2. :wink:

Robin2
Posts: 180
Joined: Sat 17 Jan 2015, 18:17

Re: Do we really need a NEXT PUPPY ?

#3 Post by Robin2 »

rcrsn51 wrote:
Robin2 wrote:But, apart from that, I don't think we should be frogmarched into new versions.
That's an odd phrase. No one is ever forced into a new version. Your old version will not spontaneously self-destruct when the new one appears.
What you say is technically correct but in practice, with Xubuntu for example, there are updates/upgrades of something or other every few days. There is never an explanation of what the changes will be so that you could make an informed decision. And if you ignore the updates for a month or so there are so many that the only practical option is either to select them all or get further out of date.

I understand completely what you say about new hardware - but why not design the system so there is a simple option to upgrade the piece you need only - for example upgrade the kernel without needing to upgrade LibreOffice and vice versa.

...R

User avatar
mikeb
Posts: 11297
Joined: Thu 23 Nov 2006, 13:56

#4 Post by mikeb »

Or build the driver on the kernel in use...that easy option seems to get forgotten...not like you buy a new dongle every week is it. :)

Well the main complaint would be that the masses would be starved of their weekly dose of freebies from the FOSS world....there would be a riot.

I find there is more software pressure...centred around the web.... most other stuff just does the job and keeps doing it.... physics and such.

Puppy 2 methods perhaps...that would be worth hanging onto...just keep the software happy with updates and yes even microsoft tell you what's in the updates if you want to know without jumping through too many hoops...which is nice because i turn them off and only have those really needed...which is a surprisingly small amount.

Fortunately I have never subjected myself to the mainstrean distro contant update boogie.... it would seem too much like vista and that would be the end of it. Plus often its not just updates but 'lets change half the system basics and breaks everything from 2 weeks ago.... unprofessional stuff unfortunately which keeps potential mainstream users away. Check what versions of linux your pet servers are running and you may notice that they are surprisingly conservative in what kernels and userspace they are running...because the want to KEEP running.

One last thing.... having new features to offer simply gets more perceived brownie points than bug fixing...the latter is classed as dull far too often. Not the way of true engineers....

ok time to duck ....... :D

mike

User avatar
rcrsn51
Posts: 13096
Joined: Tue 05 Sep 2006, 13:50
Location: Stratford, Ontario

#5 Post by rcrsn51 »

mikeb wrote:Or build the driver on the kernel in use...that easy option seems to get forgotten
Easy for whom? Consider the huge amount of work here to keep a Puppy version current with hardware.

User avatar
mikeb
Posts: 11297
Joined: Thu 23 Nov 2006, 13:56

#6 Post by mikeb »

new device appears...build driver... OR configure new kernel, build, adapt initrd to modules and latest quirks, alter system to deal with latest quirks, update iso and start all over again with bug fixing...sounds fun.

Perhaps it is easier from a users point of view to build a one off driver but I am sure some of the above would need effort from a dev or two.
Perhaps also I have found driver building to take little longer than the time it takes to unpack the source...especially when working with a familiar kernel.

Also another perhaps is I do notice that since microsoft rarely alter their kernels, driver maintainance is a relatively simple task which the venders themselves happily undertake. Last big headache was the introduction of the vista and newer model....still NT just additional security stuff to handle.

Back to the first point.... every new puppy means tempestuous has to start all over again with another... drivers for ... thread...along with the change of toolkit and familiarization required with every kernel change.

Constant change is not necessarily a good breeding ground for creativity and progress.

mike

User avatar
rcrsn51
Posts: 13096
Joined: Tue 05 Sep 2006, 13:50
Location: Stratford, Ontario

#7 Post by rcrsn51 »

mikeb wrote:every new puppy means tempestuous has to start all over again with another... drivers for ... thread...along with the change of toolkit and familiarization required with every kernel change.
Please give an example.

Maybe there's a reason why recent 3.x kernels don't need a new "drivers for" thread.

User avatar
mikeb
Posts: 11297
Joined: Thu 23 Nov 2006, 13:56

#8 Post by mikeb »

At some point there will be...lucid did not have one initially...... or is the current idea to obsolete a release every month which sort of brings us back to the constant update cycle which the op referred to.

In the case of puppy a kernel change is not exactly as straight forward as it is in say ubuntu either.

I also curious how its now become the case that a set of drivers are associated with a particular kernel ... ...for example I changed alsa on puppy 4.12.... was a quick switch and made a netbook happy....kernel remained the same keeping everything else happy. Though they need kernel approval many drivers are developed independently of the kernel and a (kernel) build is simply contains a snapshot. Your pet wifi dongle might have a fix for it 2 days after you implant the latest (kernel).

If the kernel is to become that monolithic then puppy had better find a better way of changing it to keep up.

What about stability.... you might get a new driver but is this weeks kernel a stable one since its trying out new (kernel) features?
Some kernel releases I had used had bad problems with wifi and NFS for example only cured by going back a release or three.

mike

User avatar
bigpup
Posts: 13886
Joined: Sun 11 Oct 2009, 18:15
Location: S.C. USA

#9 Post by bigpup »

For the most part, Puppy has not changed much.
The basic desktop appearance and installed programs has been about the same, for some time.

The hardware support issue is always going to be a driving force for change.

What now seems to be driving Puppy version release is two things.
People wanting programs to have more and more features.
Bug fixes.

As these Puppy core programs get added features/bug fixes.
It seems to be the most common solution is to just release a new version of Puppy, with the most recent core programs.

Tahrpup 6.0 released using the latest programs in Woof CE.
Some of those programs have now been changed by adding new features/bug fixes.
So, what does the developer of Tahrpup do?
If bugs are found in Tahrpup what does the developer do?

Slacko 5.7 implemented Update Manager.
Tahrpup 6.0 implemented Tahrpup bug fixes

I can state for sure, that without the ability to update Tahrpup, with bug fixes and program improvements/bug fixes.
Tahrpup would not be as good a Puppy version as it now is.

Even Tahrpup 6.0 reached a point at which the amount of bug fixes and program improvements required a new release of the Tahrpup iso.
If for no other reason, so that a new user would get a version fully updated with the latest bug fixes, out of the box.
The things they do not tell you, are usually the clue to solving the problem.
When I was a kid I wanted to be older.... This is not what I expected :shock:
YaPI(any iso installer)

tempestuous
Posts: 5464
Joined: Fri 10 Jun 2005, 05:12
Location: Australia

#10 Post by tempestuous »

rcrsn51 is right about drivers. Modern 3.x kernels seem to keep pace with the latest devices, and there's little need to provide third-party drivers any more.
Unfortunately as new drivers appear with new kernel releases, they often require new firmware - that's another can of worms.

For example, Slacko 5.7.0 gained the brcm wifi drivers, but the corresponding firmware was not added.
Last edited by tempestuous on Mon 26 Jan 2015, 12:20, edited 1 time in total.

darry1966

Re: Do we really need a NEXT PUPPY ?

#11 Post by darry1966 »

[quote="rcrsn51"If some community members had their way, we would still be stuck on Puppy 2. :wink:[/q
over exaggeration of the century.

Robin2
Posts: 180
Joined: Sat 17 Jan 2015, 18:17

#12 Post by Robin2 »

Many interesting comments.

I wonder would it be possible to package Puppy as a pair of SFS files - one which has the kernel and other core stuff and the other with the applications.

Then one could upgrade to the new core stuff without needing to download all the applications again.

It would also make it easy to have different "flavours" - different groups of applications but a common core.

...R

darry1966

Re: Do we really need a NEXT PUPPY ?

#13 Post by darry1966 »

Robin2 wrote:OK, maybe the title is a little provocative ...

But a big part of the reason I switched from Xubuntu to Puppy was to get off the "update treadmill"

I reckon there are folks building parts for Ubuntu (and everything else including cars and clothes) who say "that's finished - what'll we do next. OH, lets make another new version or else we'll have nothing to do"

Of course there is a need, from time to time, to bring out a new version when an error is discovered.

But, apart from that, I don't think we should be frogmarched into new versions.

Let the developers of new versions advertise the improvements and then I can decide if they are of any value to me. But don't include them automatically.

...R
Thank you for your thoughts Robin2 I also agree that using Puppy has stopped me having to be stuck in the update treadmill.

There are great Pups like Lucid which still have life left in them yet.

Mikeb has a good point that drivers can be written for these older kernels eg. RTL8192CU works really well in 412 and 4.31 thanks to Mr Tempestuous and others.

Everyone has Pups which suit them for me its Lucid, 3.01 and 4.12/4.21/4.31 JRB Version. I even like Carolite 1.2.

There is a place I suppose for old and new but even kernel 3
versions will need extra driver threads one day or do we simply throw them away. Sorry I'm personally not joining the treadmill.

Lastly Robin2 welcome and enjoy your Puppy journey.

User avatar
mikeb
Posts: 11297
Joined: Thu 23 Nov 2006, 13:56

#14 Post by mikeb »

Modern 3.x kernels seem to keep pace with the latest devices, and there's little need to provide third-party drivers any more.
Could that be in part due to a slow down in the appearance of new hardware with as you say the exception of the firmware requirements. There was after all the boom associated with the vista era which was intented to obsolete all the XP and older gear and associated systems and in linux land keeping up with this weeks hardware baby was indeed a challenge?
The stability comment still stands in my book...sometimes new kernel features can cause problems...which is what new releases are about rather than new /updated drivers..that just happens because they happen to be the versions around at build/release time.

Well if you are having an easier time then thats a bonus tempestuous :)
I wonder would it be possible to package Puppy as a pair of SFS files - one which has the kernel and other core stuff and the other with the applications.
I did this with a slax kernel change ...was trying lucid with slax 6...and yes made sense...for puppy though (and slax) as it stands you also have to have the initrd as a matching pair...not the worst deal I think...otherwise all initrd drivers need building in or use perhaps the humongous initrd approach.
Also as it is that extra sfs would not be handled very gracefully....some script changes needed but again no biggie. After all getting the layering correct is only a minor adjustment as is loading more than 6 sfs.

mike

rokytnji
Posts: 2262
Joined: Tue 20 Jan 2009, 15:54

#15 Post by rokytnji »

Having 6 dogs. And 2 cats.

Image

One can never have too many Puppys.

When faced with order or chaos. My druthers always goes with chaos. 8)

anewuser
Posts: 93
Joined: Sun 05 Feb 2012, 20:00

Re: Do we really need a NEXT PUPPY ?

#16 Post by anewuser »

Robin2 wrote: But a big part of the reason I switched from Xubuntu to Puppy was to get off the "update treadmill"
.
.
.
Of course there is a need, from time to time, to bring out a new version when an error is discovered.
Depends on use case, and enthusiast/hobby need. For most this is not a project about customers/users/$$$. It's more about trying new script hacking/programming ideas, mods and custom libs improvements. These make the 'update trend' I think.

For others it's about getting old hardware to function again, simple and fast. These only need stability, and eventual spare apart updates. Gotta love all those motherboards work again.

With that said, for instance, I'm currently in need of office applications on wary 5.0 or racy5.5 on sfs form because abiword isn't cutting it anymore for that I will probably try lazy puppy on the future while keeping customizing my puppy to my needs.

puppy is the best for cheap used netbooks compared to windows7 default installations that they came up with (using a intel gma3150 motherboard, also known toshiba n455 aka if you need to test something on this let me know)

I also want the newer quirky builds to try f2fs and not wear down my usb drives. That's only possible with a 3.8 linux kernel, which is not on wary/racy or lazy puppy I think. So a new excuse to get me a new puppy hehe

Puppy is all around pretty stable, and great for satisfying particular needs.

PS: I'd also try a https://www.jolicloud.com/jolios netbook/laptop based puppy. For instance, I can't seem to adjust screen brightness, volume seems pretty low compared to windows7's settings -default os now borked. But laptop based specifically to improve battery life reporting and customizations. Also some keyboards keys fixed on windows7 don't work as expected or at all on puppy -functions keys-

User avatar
mikeb
Posts: 11297
Joined: Thu 23 Nov 2006, 13:56

#17 Post by mikeb »

...yes the hobby distro...I forget that at times. I get far too serious and thats simply a bad idea on here :)

Its handy for experimenting on.... not used it as default system since puppy 2.12 (which as it happens was the main testbed) .

F2fs... hhhmm...I think there are better ways reducing flash wear ...if that still applies with newer devices...eg loading all to ram including saves...then the flash stick can be removed...cant get less wear and tear than that and no fancy file systems needed either...indeed the aforementioned puppy 2 can do it as a variation of the multisession approach....been using it for years.

Office suite...well softmaker is a good one if it fulfills your needs...otherwise its libreoffice time. Packages for both abound here.

mike

chillinfart
Posts: 88
Joined: Mon 22 May 2006, 18:43

#18 Post by chillinfart »

I started with 1.x versions (2006) and when i changed to 4.0 on my dead PIII, that was a big leap. During years Puppy took it's own way to fix bugs or add apps, without need of take repositories from others as happens today.

I just changed to Slacko (2011-2012) when Wary and LuPu was too old for my netbook, specially wifi and video drivers (F+++ Intel). Xorg server did too a big leap. I remember when puppy even had lack of a devx package and it should be remastered on iso before sfs implementation, Xorg also needed a separate entry (DRI and Mesa drivers).

And now i changed Slacko in favor of Lighthouse testing built-in video drivers and bundled packages (works better on my netbook and desktop system). Now i hope that Fathouse gets mature.

That movements took a lot of time, at least one year, the recommended time to reinstall Windows. So isn't surprising for me to see why people moves to a fresh new iso every year.

From mikeb thoughts, working on a certain kernel gives more life than moving to a new one. However, puppy projects are moving currently to get specific goals and when they want to run, leaves behind some needs.

i.e. Intel Atom still being a headache to optimize, when i tried to compile intel drivers on Slacko's kernel 3.10.2 i got a slow boot and the need to change parameters on xorg.conf to make it stable. And lastest intel drivers requirements are too far from Slacko and Slackware 14 (Mesalib 9.2.5 can boost 3D performance on N450 family, but i had no luck to compile it without gaps).

Yeah, it's a dilemma.

Atle
Posts: 596
Joined: Wed 19 Nov 2008, 12:38
Location: Oslo, Norway
Contact:

#19 Post by Atle »

In my understanding, Puppy covers basically 3 major areas of hardware

64 bits
32 bits with or without PAE
486/old machines

Assuming that is right, that is the 3 "families".

32bits is for sure the larger family and the most used, like Slacko and Lupu.

Then there is the older stuff like Turbopup, and 1,2 and 3 versions that still does a good job on really low ram/CPU.

As for the 64 bits, FatDog really rules and there are others and new ones in the pipeline.

What I miss is that all the great ideas hidden in the forum is implemented in maybe at least one version of each category.

I have seen so much great software written, only to die out somewhere hidden in the forum.

I can give examples:

Shareinternet(Puppy as a router)

Hostapd(Puppy as a access point to share internet, gaming or webserver)

LAMP(A standard good server based maybe on lightweight Hiawhata in stead of Apache)

I bet there are lots more to be mentioned.

Just to give some more examples, easy ways to:

Delete data properly(proper disk wipe)
All categories with the .exe installer

Just some thoughts


atle

Robin2
Posts: 180
Joined: Sat 17 Jan 2015, 18:17

#20 Post by Robin2 »

I've been thinking more about all this and taking account of various comments.

I reckon I got off on the wrong track with Puppy - I thought I caught a glimpse of the Holy Grail.

When I saw that there was a "simple" way of building distros with Woof as well as the opportunity to convert an existing setup into an SFS I imagined that it may be possible to escape permanently from the update treadmill.

Two things have changed my mind.

First, even if I did take the time to figure out Woof it would be completely impractical to understand the differences between different kernel versions and cross reference those differences to the different versions of application programs and the features in those programs that I like or dislike or need or don't need. In other words I would not have the knowledge to build a better distro than any of the standard distros.

Second neither I nor AFAIK any of the Puppy community has any control over the onward progress of the Linux kernel developers and they seem to have scant regard for long term backwards compatibility - probably on the basis that it is not too difficult for an application developer to build a new version to go with a new kernel.

On the other hand Puppy has shown me that it is very practical to try out the next version in such a way that I could easily revert to the previous version if I wanted to. Whereas the normal Xubuntu upgrade process is very much one-way.

Of course, having figured out how easy it is to install different versions of Puppy in different partitions on my hard drive I have come to realize that I could do the same thing with Xubuntu - the partitions would just need to be a little bigger.

...R

Post Reply