How about an LTS (Long Term Support) stable Puppy version?

[darkcity]

great idea, maybe its a simple adding some information to the thread title in the 'Bugs' section and adding sub forums for each version off Puppy (with an 'Other' forum for less popular derivatives).

I know we already use [solved], maybe using [confirmed] would be useful too. Possibly a [Feature request] option would be handy.

One problem is that only mods and original posters can change the title - maybe some extra mods allowed to change titles just in that sub forum?

[SilverPuppy]

If I were going to pick one, it would be Lucid Puppy. It supports virtually all modern hardware and a lot of older hardware. I am posting this from a Compaq Evo N410c with a 1.2GHz PIII processor and a 16MB ATI something-or-other video chipset. Puppy is installed to the hard drive (full install) and it runs perfectly. It also supports installing packages from the highly-popular Ubuntu repositories, and standalone .DEB packages, as well as the excellent .PET format. It has an attractive, clean, polished appearance, and an excellent feature set.

The fact that 5.2.8 is in its fourth bugfix update says that this is already recognized by the community. Why not make Lucid Puppy 5.2 the official LTS version of Puppy, and put nothing but stability development into that version? Focus the effort on perfection, and then as feature ideas come along, they can be added to a new Puppy, to be kept in alpha for a long time, then seriously tested as a beta for awhile, then finally released as stable several years down the road, once it has achieved the reliability of Lucid, and making it the new LTS version.

One of the great problems with Linux is the "herding cats" that is Linux development. Puppy seems to be the one distribution that could break the mold and really make a significant dent in MicroStupid's market share. I have been around Puppy for awhile, even contributing a tad bit of code, and have seen what a great community this is. The danger is in the last post: "Feature requests." In order for any product to become a LTS product, all feature requests MUST be deferred to a future version rather than shoved into the current, or you can never achieve a truly finished product, since new code and therefore new issues are constantly being introduced. Most of the cat herd hasn't realized this, or at least, hasn't developed the discipline to do this. Can we do this?

I think that Puppy with a LTS product would seriously begin to eat the lunch of some of the older, more famous Distros. I will post more about that soon. Look for a new forum entitled "Confessions of a Puppy cheater." I will start it basically with a blog post about my recent experiences with some *GASP* NON-PUPPY Linux! ;D

[SilverPuppy]

Here are a few things that I think should really be fixed in Lucid to make it friendlier for normal people:

1. The improper shutdown code that I contributed wasn't really totally polished, and wasn't used intact. It doesn't work at all in its present implementation. Because the system assumes UST on the RTC at boot, the superblock last write time is often in the future if you live West of Greenwich, which makes a quiet fsck fail. Either the clock needs to be set first, or I don't know what, but it doesn't work. I also had set it up to reboot after the fsck so it never dumped out at the # prompt, which is anathema to "normal people." This has been removed or broken in the current implementation, and I have yet to figure out which.

Enh.......I forget the rest. I'll post more later, perhaps in the right place. This gives a good example of a bug to squish.

[SilverPuppy]

I remembered another. I suppose this falls under new feature, but it is such a glaring problem that I think it should fall under the category of finishing an existing feature that only partially works.

When the Package manager installs a package, any files that already exist are overwritten. If this package is removed, they are then deleted, and nothing remains. This has crashed my install on at least one occasion. The package manager install script needs to make a list of files in the package, see if they already exist, and if they do, it needs to archive them with the uninstall logs. The uninstall script needs to then be modified to not only remove the new files, but also put the old ones back.

This is not a perfect solution, as it could open some interesting versioning cans of worms, but it would be a great improvement over leaving missing files upon uninstalls. Perhaps I should pursue this as its own thread.....I've toyed with doing it myself, but I'm not really that clever in bash, and someone who was could do it a lot more quickly, and probably better, too....

[jrb]

When the Package manager installs a package, any files that already exist are overwritten. If this package is removed, they are then deleted, and nothing remains.

You must have a full install?

When you use frugal, livecd or usb installs no original files are ever deleted as they are contained in the puppy.sfs read-only file. This is the main reason I quit using full installs early on in my puppy career. I love the fact that no matter how badly I screw up the original operating system is intact and usable.

Forget full-installs. They are only for REALLY ram challenged machines.

Cheers, J

[tallboy]

Hi SilverPuppy.
While it may be a sidetrack, and not belong in this discussion, there is an interesting, but seemingly not actively developed, distro named GoboLinux, that has an entirely different approach to solve the package management problem.

An alternative distribution which redefines the file system hierarchy by installing everything belonging to one application in one folder under /Programs, and using symlinks from /System and its subfolders to point to the proper files.

It may not be very attractive to some; I seem to remember some early puppy versions with lots of symlinks, and also the eagerness to steer away from that path!

It may nevertheless be some ideas in GoboLinux worth looking at, the possibility to run several versions of a program simultaneously is quite intriguing, and it provides a solution to running programs based on several different distributions.
More info: http://en.wikipedia.org/wiki/GoboLinux for an overview, and of course http://gobolinux.org/

tallboy

[SilverPuppy]

The symlink approach was taken by M$ in the Longhorn family of Windows to resolve that issue. In Vista, "7" (Vista SE) and "8" (Vista III) they use symlinks in all the Windows folders, and the actual files live in a very messy directory called winsxs. The winsxs folder gets huge, cluttered, and is easily broken, but I agree that the modified approach taken by GoboLinux might have some merit. However, that is exactly the kind of thing that I was talking about with "herding cats." That is not a bug fix; that would be a new feature. It should be directed at a new LTS version, not the current one. It would introduce all kinds of new issues with the installer that could take a long time to really shake out properly.

See how much discipline a LTS version would take? I still think it would be worth it.

[darkcity]

I agree that a polished version of Puppy with everything working would be great. However all the developers are working on a voluntary basis, and therefore, correctly only work on what interests them. There is no Puppy foundation no one receives any money, Barry may get a few donations here and there. Implementing new ideas is more motivating for many people than chasing bugs.

don't if you've seen 'how the project is run'-
http://puppylinux.com/development/project-statement.htm

there is a thread on package management here, might be worth adding GoboLinux to the discussion-
http://murga-linux.com/puppy/viewtopic.php?t=75383

on frugal install the PPM may mark a file as deleted (in the PupSave file) even if it is physically there in a read-only SFS

[SilverPuppy]

Could someone START the Puppy foundation? It could be a community-based organization that develops a single version of Puppy Linux at a time, much the way M$ releases a single version of Windoze at a time, and then works on it to perfection. It would have its own foundation website with its own bug trackers and feature request links, its own package repository, and would follow the plan I have laid out for development cycle. It could be funded by a combination of donations and paid-for technical support, which could be made available if someone was willing. Once the product reached its LTS release, a mandatory minimum donation (with option to donate more) to download said product could be implemented, and retail discs could be made available at a fixed price. I'm not sure how the GNU GPL works with things like that, however. Red Hat Enterprise Linux is a purchased product, and I'm not sure how they get that all legal, but there must be a way! Developers could be paid as independent contractors on a per-bug or per-feature basis until the time came to have a staff team, at which point per-bug cash bounties could still be offered to incentivise community bug fixes.

I guess with Red Hat raking in a Billion dollars last FY, I see a market for a similar consumer product, supported by a development team that is financially tied to the product, to provide an easy-to-use, feature-rich Linux distribution that is truly accessible to the masses, and MAYBE, just maybe........finally eat Bill's lunch.

[darkcity]

its possible but it would be a fork of Puppy . . .

...Anybody is welcome to fork the project and manage it anyway they like. The thing is though, it isn't easy, an incredible amount of commitment is required. If you like Puppy but would like to see more formal management, ...

http://puppylinux.com/development/project-statement.htm

[jemimah]

Could someone START the Puppy foundation? It could be a community-based organization that develops a single version of Puppy Linux at a time, much the way M$ releases a single version of Windoze at a time, and then works on it to perfection. It would have its own foundation website with its own bug trackers and feature request links, its own package repository, and would follow the plan I have laid out for development cycle. It could be funded by a combination of donations and paid-for technical support, which could be made available if someone was willing. Once the product reached its LTS release, a mandatory minimum donation (with option to donate more) to download said product could be implemented, and retail discs could be made available at a fixed price. I'm not sure how the GNU GPL works with things like that, however. Red Hat Enterprise Linux is a purchased product, and I'm not sure how they get that all legal, but there must be a way! Developers could be paid as independent contractors on a per-bug or per-feature basis until the time came to have a staff team, at which point per-bug cash bounties could still be offered to incentivise community bug fixes.

I guess with Red Hat raking in a Billion dollars last FY, I see a market for a similar consumer product, supported by a development team that is financially tied to the product, to provide an easy-to-use, feature-rich Linux distribution that is truly accessible to the masses, and MAYBE, just maybe........finally eat Bill's lunch.

Sure, go ahead and start it.

[SilverPuppy]

I guess I sort of figured that would be the response. Truth to tell, I am considering it quite seriously. I'm not the best Linux coder ever.....in fact I'm marginal at best, but Bill Gates wasn't the world's best programmer, either....he was a visionary and a business man, and he knew that with those skills, he could find the coders.

Maybe I will. I am going to start a new thread probing the feasibility of such a project.

[jemimah]

The problem is paying the coders... geeks generally only implement someone else's ideas for free if they worship the ground that person walks on.

I think you are seriously underestimating the amount of capital needed.

From a 2004 article about redhat:

For the second quarter of its 2004 fiscal year, the leading Linux company posted a profit of $3.3 million, or 2 cents per share, on roughly $29 million in revenue.

That compares with a profit of $1.5 million during the previous fiscal quarter, and a loss of $1.9 million for the comparable quarter a year ago. Revenue for the second quarter of fiscal year 2004 increased 6 percent from the prior quarter and year-over-year by 36 percent, the company said on Thursday.

The company's second-quarter net operating profit was $240,000, compared with a net operating loss of $1.1 million in the prior quarter and a net operating loss of $4.7 million during the comparable quarter a year ago.

Keep in mind redhat was founded in 1993. It took nearly 10 years and millions and millions of dollars to make it happen.

[MinHundHettePerro]

I guess I sort of figured that would be the response. Truth to tell, I am considering it quite seriously. I'm not the best Linux coder ever.....in fact I'm marginal at best, but Bill Gates wasn't the world's best programmer, either....he was a visionary and a business man, and he knew that with those skills, he could find the coders.

Maybe I will. I am going to start a new thread probing the feasibility of such a project.

Might initially be a success ..........

After some time I guess the name SilverPuppy would be spelled with an initial $ ............
/MHHP

[SilverPuppy]

Well, more feedback would be appreciated at my new thread. http://murga-linux.com/puppy/viewtopic. ... 288#616288 I did copy from this a little, but take a look.

One thought about 1993 and Red Hat: in 1993, M$ was cool. In 2012, large segments of the PC market are disgusted with M$ and looking for something else. Linux is being heavily "checked out" right now because it isn't M$, but the lack of a similar product in terms of support and usability is causing most to turn back to M$ in despair. Properly developed and marketed, it could attract investors and talent from far and wide, and become a real force in the market.

Maybe I'm nuts. Maybe we'll give it a whack and see...............

[jemimah]

If you want something like this to succeed you need to pick a specific hardware platform. It's simply not possible to get "just works" on all generic hardware with linux.

But if you pick something specific, you can make it work better than windows or whatever.

So the hardware you choose needs to be extremely appealing: ie, sexy form factor, and cheap price. Bonus points if it can't run windows very well or at all.

Think about the original EeePC.

Also the OpenPandora might be a good example, but I think it is still vaporware.

ChromeOS and the google chromebook would be a third example.

[simargl8]

LTS (Long Term Support) stable Puppy version is a good suggestion.

[starhawk]

Wary is our LTS Pup

[nubc]

I would say commercial production of disks is a waste of time and resources. Wary is a good choice for LTS, but it's a lousy name, promo-wise. The demand for desktop computers is greatly diminished in recent years, and to a lesser degree, the same for laptops. If you have a plan for promotion of Puppy, let's not be grandiose with objectives. We'd be lucky if a stable LTS version raised Puppy a couple places at DistroWatch. Don't forget the very controversial security issues re running as root. This is bound to be a big distraction from Puppy's merits.