how is it possible to offer something to "everyone" at this point, and does such an idea even matter?
a number of people have complained about splintering lately, even a lack of unity. could a puplet help bring people together? obviously, a distro never brings EVERYONE together. splintering happens... diversity is even a good thing. but there is a point of balance. too much coalescence kills- the windows monopoly is such a victim, moreso its users. likewise, and on the opposite hand, too much splintering kills.
puppy continues to strive for quality, but the splintering continues. is puppy really meeting everyone's needs? can it be made to meet more needs without losing what originally made it exceptional?
it's my feeling that a puppy can be built to bring more people together, to fit not just a larger group but larger variety of needs, without abandoning its original strengths and specialties. but first we would have to examine those:
* puppy is small, and fast.
it needs very little ram for everything it does.
it works great on older machines!
these are things we could lose as fast as windows has lost them.
* puppy is easy to use... it's not designed foremost for source hackers.
source hackers are great. we have some. they are of great help.
if you want to hack source, puppy will let you.
it's easier to get gcc working in puppy than it is in ubuntu... unless you don't mind its package management.
puppy has/had one of the most straightforward package management systems ever.
through a lot of tinkering, we've managed to add features to it, but it's been a while since it was as reliable.
people frequently have to download .pup and .pet files the old fashioned way to get them to download
and install properly. soon i suppose we will be back on track.
puppy has wizards that continue to improve in quality and ease of use. should this go without saying?
absolutely not. this is part of what makes puppy puppy.
* puppy is flexible. it's tiny, but you can bloat it up with ooo and kde. sfs is a wonderful way to fit loads of stuff into tinier iso's and packages. it isn't aimed JUST at n00bs or JUST at peopel that want full control (root, FFS!) but it can cater to both in a way that might be unprecidented.
can it go further than that? yes, i think so. for one thing, we can stop making puppy bigger. we can stop breaking things without fixing them. we can get a LITTLE more organized that we used to be and have it make a LOT of difference. we can stop puppy from becoming 90mb, 100mb, 200mb, 600mb. dvd like EVERYONE ELSE does. the new graphpup alpha is 100mb
this is terrible considering that grafpup mastered sfs, that it found a line to draw in the sand: 400mb and tried to stay under it, BUT ALSO with a core that was under 100mb.Three Digits is Too Many!
to be fair, 100mb alpha doesn't mean 100mb beta. grafpup 2 (otherwise, a shining project worth watching closely,) might not break the 2-digit barrier. you can have a useful puppy-like 80mb core os, and then optional SFS for things like ooo and kde. in fact. as anti-bloat as i am, i think the kde sfs is hope for puppy. puppy can go on and off the tiny track... big fat puppy, little puppy, big fat puppy, little puppy. sfs is a real-live-working body-builder suit for a chihuahua. and the 80mb grafpup 104 is much like puppy 1.07... with better wizards and easier (more established) methods to get into kde and ooo. just add abiword (it has ted) and mp. maybe nvu (not to the core!)
**** where to begin?
first, you have two kernels. you have the 2.4 series and the 2.6 series. the 2.4 series is timetested, fast, less likely to crash, has a gaping security hole in it that is patched in 2.6, and works more reliably with pcmcia cards in laptops and with cdrecord than 2.6 does.
wow! lots of differences. add to this that the 2.6 series belongs to "puppy 2," and that it is based on the neat t2 project, and that puppy 2 has a lot of things that puppy 1 didn't. does it all rely on the kernel? NO. a lot of it relies on development.
one thing that would be very cool to see (so far no one has made a case for it) is a puppy that you can MORE OR LESS swap out between one kernel and the other. SO MUCH of puppy works from puppy 1 to puppy 2... not all of it! but so much.
this could be a GOAL of puppy, and the advantage is that people aren't arguing about puppy 1 and puppy 2. is it worth the trouble? that depends on how far you take it vs. how much trouble it is to take it that far. i think MOST of puppy could be interchangable, if it was slightly more organized without becoming too complicated.
**** lgpl coming soon?
another exciting development in puppy is becoming more officially gpl. i used to make a stink about this, proportional to my concern. i have NEVER once accused puppy of VIOLATING the gpl. i know better... there was a guy that used to say that about puppy, but barry has always gone to the trouble of making sure puppy doesn't violate anything.
still, it's hard to make a gpl puppy. there are scripts that are copyrighted. THIS HAS BEEN FURTHER CLARIFIED. in fact, there is an faq that says in essence, the only thing that is still protected (i.e. the only part of puppy you still need barry's blessing to make use of) is the Trademark- "puppy" this may seem quite odd to many of us merkins. in the u.s. a trademark's hold changes when you change half of it. puppylinux and puppytelco are not an issue in the usa. in au or the uk, however, it may not fly. it makes some sense to honor international and aussie restrictions, as puppy is an international and aussie product. but this leaves you with an entire os to play with in the spirit of foss.
BUT A LOT of people are still going to want it to be officially released as gpl or lgpl, which barry has talked about doing in the near future. note that what really makes puppy puppy are the scripts that puppy uses to do things like package management and boot, and loop around and (this is really unique, find another live cd distro that does this) let you remove the live cd after booting. (i've seen a floppy distro that does, but not a live cd.)
*****barry is likely to release those scripts as lgpl in puppy 2.14 or later.***** what's the advantage of this? well, it's less confusing than trying to find the faq (keep in mind how many puppy websites and starting points there are... i looked for months for this information) for license info. most people look in the distro itself, where older, more confusing information is listed.
the point? when those scripts become lgpl, it will be a huge advantage for puppy. it will be a reason to use the latest version, or at least the startup scripts from it, so that real L/GPL derivatives can be created. this is a point for puppy 2. it's also a point for making puppy 1 and puppy 2 more interchangable. gpl means something can be built to last, and this is an important feature to keep in mind for all puppy derivatives. watch for the lgpl scripts, and grab them and use them.
**** other gpl efforts
of course, few people in the puppy community are concerned about this. it's a larger concern outside the puppy community, in the much larger linux community. if you want a bigger audience, which many people seem to strive for, gpl can help. ubuntu has perhaps taken it farther than is sane, trying to cut out anything that is patented. patents don't have an equal hold in all countries, and some countries are going so far as to try to ignore or nullify software patents outright.
considering that honoring *all* software patents means that you can't make a window with TABS in it, i'd say it can only be taken so seriously. gpl is another matter, relevent not to trademarks, not to patents (just yet) but ALMOST exclusively to copyright. it is a driving force behind linux development, and what keeps linux libre while allowing businesses to embrace it.
for those that have a desire to demonstrate gpl further, it's been suggested for another distro that any of puppy's non-gpl tools (puppy has a few, not a problem nor a violation there!) be put in a "nongpl.sfs" this can be included in the os core, the main iso, this is not a problem. it's a way of making it EASY to switch puppy between all gpl and mostly gpl. worth a thought: nongpl.sfs.
**** reliable releases, sane schedules
puppy started as a hobby and an obsession, and it should remain a hobby and an obsession. on the other hand, it's much more difficult to manage a group of people creating puppy linux, and that's exactly the task that creating puppy has become. and the idea of one wreckless genius pulling all the strings has ceased to be puppy's model. barry has talked about moving past puppy, leaving it in the hands of others. this is the fate of all free open source software projects, that or oblivion. barry is wise to look for people to leave in charge of the official, original project.
is it possible to have the level coordination now, that once was possible in the days when puppy was a 1 or two man project? the simple answer is no, that will never be possible again, except when "puppy" is a "puplet," a relatively small side project. even then, few men on earth have barry's knack for tapping out an entire new version in weeks, testing it in a few hours, and having something useful come out of the blitz.
that's okay!
"lesser" developers (or at least, calmer ones) are forced to do old-fashioned things that aren't as much fun... like label things carefully. "every version is really a beta" doesn't work if you are carrying something forward that will truly be friendly to n00bs and not have people scrambling to patch and fix things (such as usb support in 2.13 that worked better in 2.12 for some people) with things like modprobe on flash sticks.
better testing means better bugreporting means better bugfixing means a better distro.
so how far do we go? well... it takes longer, but i'm going to say more than a week for alpha, more than a week for beta, and more than a week between beta and release.
also: "alpha" means that you're still planning things, but that you have a plan (not having a plan yet is pre alpha... okay, even this is relatively casual... lump that in with alpha)
"beta" means that you're trying VERY MUCH to not add new features, make big changes, just bug fixes and (where applicable) choosing which release.
"release" means... NOT BETA.
it's really very simple, it's a matter of taking the process seriously and having it as a real goal. it doesn't mean you have to run around with a clipboard or a pen behind the ear, or hire a cpa or even setup a CVS repository. puppy can keep its spirit and can be labeled better... and run at a pace that lets the users catch up.
another advantage is that the "release" would only come out ever 2 to 3 months or so. or who knows? 2 to 4 releases a year! imagine how well puppy could be Documented if it slowed down just a bit?
**** watch the fringes
what are critics saying? not people that never did like puppy, (although they're worth a glance and a half at least!) but what about people that used to love and swear by it? are they suddenly unhappy? can anything be done?
despite the spectre of pleasing everyone that complains, such complaints are priceless clues for development. some you can argue, some you can't, some you must, and some you shouldn't. when puppy starts to splinter in many directions at once, you cannot ignore the possibility that puppy is losing some, evne too much of what made it great. try to keep as much of it there... if you can advance it without losing how fast and small it is, great! and if you want to make your own bloated puppy... well, you're free to! but listen to the people on the fringes now and then. they may have the idea you will actually need later... if they don't say everything in the most tactful way, it doesn't mean they're wrong. you may find yourself agreeing later. generally both sides of an argument are both right and wrong... finding a middle ground that isn't too weak itself and aiming for the "best of both worlds" is what innovation is really all about. be afraid to lose your legacy, but don't be afraid to gain insight. or hold onto your legacy, make careful notes of what makes puppy great, and stick to that and fear nothing.
**** never have more than one or two calculators
it's just too time consuming and even confusing to people. it shouldn't be, but if people want to try five calculators they can download five calculators. or put them in a randomcrap.sfs
this goes for scripts that are vital, too. a restore script is important. never have more than two of them, or people will wonder which one is the good one, and it will be the source of lots of frustration. most people would rather use the right thing the first time.
with dialup this is another matter. ideally, it wouldn't be. it's just too hard to connect to an isp in linux, but it's pretty easy to add 2+2. pick a calculator, maybe throw in that unit converter, and leave it at that.
**** do something for those guys making barebones pups
one of the things that gave me great hope about puppy was pizza's script to remove things from puppy. having just tried ubuntu, i was extremely frustrated with how much CRAP i was saddled with that i didn't want. synaptic didn't let me add anything, everything refused to install without 5 other things, and i had to hunt them down and download them. WHEE, what a POS. couldn't STAND it.
but what else? it wouldn't let me REMOVE A-N-Y-T-H-I-N-G either! everything depended on everything else! i couldn't remove gnome because gedit needed it! i couldn't remove gedit because gnome needed it! oh, man, just LET ME DELETE SOMETHING!
so puppy actually has this. i'm just saying, it's important. cater to this need.
furthermore, don't make people on the CLI use vi. when you are forced to use the cli, either it's because you're that good or you're that frustrated. let's not forget mp, eh? i mean it's not in grafpup, i will make a .pup for it. okay, but i can't imagine it's that large. it's that important. even tho we are talking about a mostly gui distro, let's try to keep in mind that some people want the cli, and they will probably hope that there are a few tools they can already use there. two things that would be GREAT for the ultimate puppy: a cli version of mut (so reliable) and a cli version of dotpet. honestly, it can be done!
**** package managers
.pet could work in puppy 1 and .pup could work in puppy 2. this among other things would help transcend the idea of a puppy 1 and a puppy 2. many of the wizards from puppy 2 could work in "puppy 1" if you could swap out the kernels.
.pet needs to be stable, designed properly, and then left alone, like .pup seemed to be (and pretty much was.)
**** window managers
i found the ultimate wm perhaps... as an alternative to, not replacement for, jwm.
i mean alternative to, jwm it's better looking than icewm and jwm, but it's smaller than the others. it comes with grafpup 2 alpha. it's GORGEOUS. and simple. no bs. is it fast enough? it feels faster than ice does.
the ultimate puppy would have jwm by default... and then that (i dont know the name) builtin (sfs maybe) as something you can switch to like yo ucan switch to and from flux in dsl.
jwm is notable above others in functionality as you can right click Kill process so much fun. ice doesnt have that?
xp and vista themes should never be installed by default. there will be dotpets for them both. ADD FUN... don't force windows looks on people.
**** browsers
there are so many new options for browsers. seamonkey or opera could go in an .sfs and something lighter could be part of the puppy core. browsers worth noting: hv3, iceweasel (like firefox), firefox, opera. there are still others. people that like bloat will LOVE sfs and people that like *lightweight* will love the new old/old new puppy!
**** gaim and seamonkey
worth consideration when you're cutting bloat out of puppy is cutting out seamonkey and gaim. xchat will let someone get on irc... anyone that wants aim, msn, yahoo have other options. make them .pup or .pet? well... it needs some libs from seamonkey. so maybe have them in a .pup together, or maybe have a standalone .pup for gaim with the needed libs. and ditto for seamonkey.
**** puppybasic
theres one thing i'm adamant about. when i got mu's 12mb puplet, i checked for something to see if it was there. it was, i was overjoyed- puppybasic. if it's there, people can make all kinds of tiny tight nice apps. if it isn't, none of them work.
it's a small thing, it makes many other small things work... you can build half a tiny os around it and cut out large bloaty apps. so would that be a dotpup like everything else? or... i mean if you want to make a puplet that is Not barebones preferably, still want to be small, but want to be unique.
a medium puppy gives them that. a bloated puppy isn't a good enough, but puppybasic is a great way to support tiny, simple, featured apps and wizards, keep puppy small, and friendly. puppybasic is about as important to puppy as bash is. even if you're not "using it," you need it.
**** summary and additional notes
the ultimate puppy would be small, fast, and easy to use. it would work on old computers. it would enable bloaty features for new computers via sfs, the way grafpup does.
it could use the puppy 1 kernel or the puppy 2 kernel. the one based on the lgpl scripts that may come soon might go so far as to put all non gpl stuff in a nongpl.sfs
the core will be more essential than the bloat. the bloat will be optional, the core will be the first focus, after quality itself. the core will stay under 100mb always, and attempt to stay under 80 (that means 75 is a GREAT job, and 70 is WONDERFUL)
it would be good to try to make .pet work in puppy 1 and .pup work in puppy 2. this among other things would help transcend the idea of a puppy 1 and a puppy 2. also, many of the wizards form puppy 2 are worth retrofitting to puppy 1, but this wouldn't be necessary if you could swap out the kernels.
jwm is worth sticking with... it's worth considering adding one more wm either to the core or to an .sfs (note that kde has a HUGE wm that you can run if you are running the kde sfs)
people really really want features. the ultimate puppy would entice people with many features but never at the cost to people that would otherwise just go barebones. look to grafpup 104 for the ultimate demonstration of balance.
few enticed > few users > few users > little support > little support > little debugging > little community > little quality.
100mb is too much. 200mb is way too much. 380 is not too much, as long as it is less than 400, and as long as the self-runnable (doesnt require other things) core is less than 100, and preferably 80-something or
less.
ideally, it's going to be easier to REMOVE apps than ever, and ideally, there will be more things for barebones puppy users... both in the gui and in the cli.
** even puppy 2.15 CE could use the basic idea that grafpup 104 has of an 80mb core and the rest in .sfs. **
worth noting: the derivative of puppy being described is the derivative of puppy that in the future is encountered by lobster, which he bases tmxxine on. unfortunately, it would take years for tmxxine to incorporate all that he found, and for copyright reasons he didn't include the lgpl scripts as they haven't been released yet. fortunately, lobster is patient, otherwise he would be more careless with paradoxes.
puppy is slower than it used to be. its size is runaway. its ease and features and eye candy are splintering in every direction. what's still important that was important before? let's document it before all it forgotten and demonstrated only by older versions.
**** what about you?
okay, so you have a puppy that is 80mb trying to be 70 or 60. you have 380mb total to work with, keeping it under 400mb with all the sfs. stuff.
what goes into 380mb? how would you sort it, if you could have puppy 1 or puppy 2 use up to 10 (maybe more) .sfs files? what would go in your core puppy? would you ever consider the space you'd have in the core to work with if you removed seamonkey once and for all, knowing it would be in one of the .sfs files? have you ever tried grafpup 104? what's so important that it should be on the main cd that MORE people use than now? your turn:
...
.
, and of course there are a few things that I take exception to a little bit. But overall thanks for the very good read.