Development

[Phill]

By the way. Zonas already started working on this. Does anyone know if he still is? I haven't been in contact with him for a while now.

Clem

I was wondering the same. He's been very quiet...

[sundayrefugee]

Ok. That sounds about as I thought. It would be an official Community Edition, subordinate to the quality controls, etc... of the Mint Team. I firmly believe that the official blessing, and having the official guidance, help, etc... of the Mint Team is the correct route.

I would be the maintainer, then. I still don't want to start unless a team can be put together But I can be the one responsible, no problem there.

Question for stable - do you mean the tools don't work in Lenny currently? I was thinking of going in that line. If so, I'll investigate some other options

Would it be possible, then, to get a post asking for interest in help in a more visible part of the forum?

Thanks!

[cmost]

I will be available to assist with this project. Might I suggest the new DreamLinux 3 (currently in late beta) as a model on which to base Debian Mint. DreamLinux has recently switched its base entirely to Debian Stable, rather than a mixture of Morphix, Debian, Kanotix and Elive. Furthermore, Dreamlinux is now built using a unique framework called "Flexiboost", making it entirely modular. It comes with both Gnome 2.20.x and XFCE 4.4 making it perfect for use with Mint's custom tools. Why reinvent the wheel? Just my thoughts.

[clem]

I could participate in this as well, not as actively as on the Main edition obviously but a bit nonetheless. In particular I would be interested in deriving a special version of mintupdate to make the jumps from one stable config of Debian testing to another. The real advantage in my opinion is to base on top of Testing with a Gnome desktop... as I did for the Alpha. The most important missing part of the puzzle here is to find/write a good installer.

Note: The current stable release of Debian is Etch. It isn't compatible with the mint tools. Testing (currently called Lenny) is. Ideally we would base on Testing of course... not necessarily on Lenny

Clem

[VaMPiRiC_CRoW]

For a distro based on Debian Stable we already have the great MEPIS.
sidux use Debian Unstable exceptionally.

So Mint Debian could be based on Testing, but this could be a problem, because the Testing could have problems and not receive a fix in a short time. I don't know how do you want to deal with this kind of problems...

I'll looking forward to this project, only to see its development and final result, because I don't like Gnome, but in the future we can see it build with KDE and others Desktop Environments...

From what I have see now, it seems that we have people that want to work hard on this project, so I hope you have all the success and bring us one more great Mint Edition...

[sundayrefugee]

For a distro based on Debian Stable we already have the great MEPIS.
sidux use Debian Unstable exceptionally.

So Mint Debian could be based on Testing, but this could be a problem, because the Testing could have problems and not receive a fix in a short time. I don't know how do you want to deal with this kind of problems...

I'll looking forward to this project, only to see its development and final result, because I don't like Gnome, but in the future we can see it build with KDE and others Desktop Environments...

From what I have see now, it seems that we have people that want to work hard on this project, so I hope you have all the success and bring us one more great Mint Edition...

Yeah, there are advantages, in a sense, that etch and sid have that testing don't. You can partially solve this with backports, partially with pinning a bit against Sid, partially against pinning against a seperate repo. Not complete solutions, of course, but these are partial.

Ok, some clarification: When I mentioned stable earlier, I meant planning for the Lenny release of stable, as I thought that would be around the time a "final" product would be about ready, anyway, so "Lenny" would be the "stable" branch it was based on, eventually. I wasn't actually ever thinking of going with "current" stable, etch. So, I'm correct in thinking that if we start going with Lenny now, we're good?

[clem]

The idea with Testing is to define a series of acceptance tests and to run these tests on a regular basis. When all tests pass it means the current state of Debian testing is suitable for the user. We take a snapshot in time of that current state.

Basically the idea is to have mintUpdate upgrade only to the latest stable snapshot... if you see what I mean. It's crystal clear in my head, I'm not sure yet how I would go on and implement this... one thing is for sure, it's hard to explain

Clem

[frank392]

but the latest snapshot is still unstable isn't it?

[cmost]

"unstable" is a relative term. It typically takes two weeks for packages to migrate from Unstable to Testing. From Testing, the packages move into Stable (which becomes the next release.) Testing is a god place to base a new distribution because the packages are cutting edge but have gone through some level of testing.

[sundayrefugee]

The idea with Testing is to define a series of acceptance tests and to run these tests on a regular basis. When all tests pass it means the current state of Debian testing is suitable for the user. We take a snapshot in time of that current state.

Basically the idea is to have mintUpdate upgrade only to the latest stable snapshot... if you see what I mean. It's crystal clear in my head, I'm not sure yet how I would go on and implement this... one thing is for sure, it's hard to explain

Clem

I'm starting to see what you mean... and I likes it.

Edit: This next bit is in response to frank, but it won't let me quote him properly

The answer to that is "sort of" A bit of an oversimplified explanation - Testing has the "big bugs", so to speak, worked out. That happens in Sid. It then gets something resembling a "bug count", which can even be seen using aptitude. When this number is acceptably low, it goes into stable. This works both for the distro and for packages. A bug might be something esoteric like a floating point multiplier off in Blender, that the odds of running into are 1 in 1,000,000. Or, it might be .iso burning not working on 700MB CD-R's in Brasero. Both are still a bug. One is more a bit of a showstopper, though.

Again, that's a much oversimplified explanation, but that's *basically* how it works. It's actually much more complicated, but that should give a good understanding.

I have often used Testing and never ran into problems. In fact, I've almost *never* run into a problem using Testing. Then again, I've avoided GNOME in Testing like the plague, because it's borkedness is notorious. And therein lies the challenge. It's why almost all Testing/Sid based distros (Kanotix, Sidux) use KDE.

That's why I look at a rather signifigant time-frame. Debian Mint isn't something that would be ready in a month, or probably even 4. However, once it was set up properly, as Clem mentioned, it *should* be rather self-sustaining, and need mostly maintainence rather than boot-strap building.

A final thought on Testing, Sid, and even experimental, though: If you've used Ubuntu or Mint, you've used all three, and probably not even known it Of course, the Ubuntu team, and then even further the Mint team, goes through GREAT pains to make sure you get a stable, "point-and-click" experience, but you've still used them

[sundayrefugee]

I will be available to assist with this project. Might I suggest the new DreamLinux 3 (currently in late beta) as a model on which to base Debian Mint. DreamLinux has recently switched its base entirely to Debian Stable, rather than a mixture of Morphix, Debian, Kanotix and Elive. Furthermore, Dreamlinux is now built using a unique framework called "Flexiboost", making it entirely modular. It comes with both Gnome 2.20.x and XFCE 4.4 making it perfect for use with Mint's custom tools. Why reinvent the wheel? Just my thoughts.

Fantastic news that you'd be able to be onboard. Can you do some recruiting? We need more if we're to start this endeavour

My thoughts - and just my thoughts - feel free to convince me otherwise.

I downloaded Dreamlinux after seeing you mention it a couple of times, and it has pros and cons. One con is that it is based on Etch, and we now know that we'd be basing on Testing with a Debian Mint

They claim to be free of Kano's work, but I noticed they still use several of his scripts

I'm not entirely sure what "Flexiboost" is, but I didn't notice any difference between it, in architecture, and any other distro.

I noticed the GNOME environment was almost completely unusable on the LiveCD, and looked pretty rough. The XFCE version (props for including two DE's on one liveCD, though) looked more polished, but as a LiveCD it was rather unusable. Most of the programs don't launch, for instance. I hope that wasn't Flexiboost.

Installing made things much better, but I found the 1 screen installer actually a little confusing, and I can actually chroot and stage 3 myself into gentoo. It's also a little buggy.

Lastly, they look to be trying to visually emulate the early Apple OS X look completely, and while that's a very beautiful look, it's not really what I have in mind for Debian Mint. They're doing it well, and although it's rough, it *is* a beta, and a very promising one. I don't propose we invent the whell, but I *do* propose we do something different There's more than one way to skin a cat. Plus, I don't think the MintMenu would look particularly attractive in that environment, and it *will* include MintMenu. And of course the most important thing is "What does the Debian Mint community want out of Debian Mint?"

The installer screen for Codecs, etc... was nicely done, but of course, being Mint, we'll include these out of the box for you. If you're a hard-line Debian purist, just install Etch. It's one hell of a distro

Obviously, you're sold on something there, so in the greatest of passionate Debian tradition, I'll expect an appropriate response (you *do* read the Debian mailing lists, don't you? )

I've actually thought of *starting* much simpler - stable snapshot and Mint tools. Build the foundation. Get it semi-stable. Build from there.

I've seen you post that you have some knowledge of installers and remastering systems. Would you like to sign on to participate in/head these jobs?

Edit: Plus, remember - this is all theoretical if we don't get proper help assembled. If it becomes the work of 2 or 3 people, the time-frame is going to increase

[Mepis Using Guest]

I am a Mepis user reading the posts and following this process as it is interesting. The thing to keep in mind is to create something that will set this edition apart from others. Stable and KDE/Gnome have been done (Mepis, Dreamlinux) and there are the unstable/testing distros (Sidux, Parsix). What will be unique about this edition that will make someone want to use it?

Anything based on repositories other than stable is not going to fly with most typical users for long. Sidux can have 100 or more updates every 2 days which is tedious to review and maintain or one and can break a system. These updates will eventually get into testing, so testing distros will have a lot of turnover too. On the other hand, anything based on stable gets too old.

Mepis has finanlly settled on stable and also creates its own repositories to keep popular packages up-to-date (openoffice, firefox, etc. and kde I think). I like that. I think it is unique and sets Mepis apart. The only thing about Mepis that keeps me looking to try others is there are other more current packages that I would like to have so a testing version would be most welcome.

Here is the question: How to tame the Debian testing beast and make a usable desktop for average users? A desktop for the users who like the simplicity of a Mepis, Mint or Ubuntu. Solving this dilemma is where this distro could be unique. Clem has stated it several times and it should not get lost in the discussion: testing updates have to be safe, a tool to prevent updates until they have been tested out as functional. Clem has also indicated that that this edition is to be based on testing continually, not on Lenny. Tesing will always be tesing while Lenny will be stable some day.

mrlag

[cmost]

@Sundayrefuge

I only suggested DreamLinux as a "base" not to emulate its look at all, rather to point to its many innovations: its modular nature and up-to-date packages (Gnome 2.20 & XFCE 4.4,) its creative use of Avant-Window-Navigator and OSX themes. Obviously a Minty Debian would have a stunning, elegant look all its own, as well as its own set of custom Mint tools. I don't believe there's anything wrong with basing Debian Mint on Debian Stable with a dash of Testing thrown in where desired. For example, Clem mentioned that the Mint tools wouldn't run on Stable due to outdated libraries. Pull in the updated libraries from Testing where needed. Compile updated Gnome. Update other packages to current versions. Maintain these in a separate repository. If the distribution is based nearly entirely on Testing then we'll be forever chasing our tails. As even the name "Testing" implies, it's a test bed for Stable; there are bound to be problems (probably more frequently than not.) Only the most experienced users will be able to cope for long. What Debian Mint will offer over all the other Debian variants is an easy to use Live-CD with an even easier installer (something like Ubuntu's ubiquity or DreamLinux's installer.) Mint Debian will provide all the tools needed to make a powerful yet easy-to-use Debian Workstation with full support for multimedia and web "out of the box." It will contain the usual array of custom Mint tools to make the system stunningly simple as well as beautiful. Best of all, it will be stable. As Debian moves from Etch to Lenny, so too will Mint. I really think it's too vast and unreasonable a taskt o expect a stable Debian Mint from pure Testing (or heaven forbid from Sid.) Just my $0.02

[sundayrefugee]

That's the thing, mate. For one, Clem would like it to perpetually use testing. That's not too far off of my idea, which was to plan specifically for Lenny,as it would be officially "stable" at around the time of release.

Where you get into trouble, with GNOME, is with a "dash of testing" You're talking about a completely different gtk version, different glib, etc... It's hard to add a program here or a library there at random, alot harder than with KDE especially.

You can partially solve that by compiling and maintaining your own versions, but it multiplies exponentially, and you have to know how to pin and tag, and watch and test constantly You have to be careful with your preferences file.

Rerolling GNOME from scratch presents even more complications. At that point, between hosting your own custom libc, gtk, etc..., you're now going to maintain your own GNOME. And debug it and tag it correctly. That's *HUGE*, and not at all as easy as it might sound You've just multiplied your problems by thousands, and now you're relying on Debian repositories for very little of your core OS You're updating from Debian very little, taking away the core concept of why Clem would like a Debian version - and 99% of the advantage of basing on Debian in the first place, and basically using an Etch kernel, your own custom-compiled GNU environment on top of that, and then trying to maintain point-and-click compatibility between that and the entire Debian repository. In a nutshell, you've just described making Ubuntu. I couldn't do that if it were my full-time, paid job

But, honestly, I don't think the *community* exists, at this point in time, to really do this project justice. For all that people may get frustrated at not being a part of the *process*, the opportunity to jump right in has presented itself right here, with only a couple of takers (bless you all, but we're not nearly enough to do and maintain it to Clem and my vision right now). I think there's a lesson in that right there

But, seriously CMOST - it sounds like you have something very particular in mind, and a nice vision. Why don't you take a stab? I'm not interested in "running the show" for an ego boost, but only because I knew, with proper resources, I could get the job done. Those resources don't exist at this point in time. When they do, I'd be more than happy to head the *official* project.

But why don't you take a stab at what you're thinking? Take the lead, and show your vision! You have no reason to be subordinate to me for any reason And I'd be happy to hack and tinker for you where I could, just PM or email me While your vision may not be an "edition", that's not to say it wouldn't be a worthwhile independant product.

As it is, unless someone more capable than myself can take the reigns, I don't think the resources exist to do this project justice quite yet

[cmost]

@Sundayrefugee While the idea of creating my own vision of Debian Mint sounds appealing on so many levels, I lack the programming skills to support such an undertaking. I would have to rely upon Clem to convert the Mint tools and craft an installer (unless I blatantly steel one from another distribution.) You've no doubt noticed something that I noticed awhile ago: Most Mint users aren't risk takers, tinkerers or willing to experiment vastly with their systems. They simply want their system to work and I don't blame them. I imagine this is why Linux Mint is based on Ubuntu - it has nearly unparalleled ease of use. Nevertheless, there are some of us who are willing to dive in and jiggle the wires under the hood, or in our case, rip out the engine and replace it with a faster sexier model! Since; frankly, it's a paradox to try to create a stable version of Debian Testing, I say let's just do the best we can and see how it turns out. Sometimes one has to break a few eggs to make an omelet. I will PM you later and we'll see what we can cobble together with our limited skills. We might surprise ourselves. In the meantime, let's query the community and get a definite "Yes" or "No" from community members who will assist with this project. I see a lot of people expressing interest but nary a peep when volunteers are requested. You and I cannot (and should not) go it alone. We're not the Bush Administration! If the people want Debian Mint, then the people will have to help to make it happen; otherwise, it won't happen, at least not by my hand.

[guest]

how about basing it on sidux??? sidux is Very Fast and Very Stable and a 1-week-time is ages in terms of their updates!

[sundayrefugee]

how about basing it on sidux??? sidux is Very Fast and Very Stable and a 1-week-time is ages in terms of their updates!

Well, there's the already-known phenomena that there is a dedicated team of Sidux techies that haven't been able to make a releaseable GNOME environment in over a year of hacking, and those guys are premium, top-notch. If they can't do it, I guarantee you we sure as heck couldn't, because nobody on the planet knows more about Sid than the Sidux team

Sidux is *THE* Sid-based distro, IMO, end-of-story, with a big qualifier - it's KDE, and for good reason - GNOME is almost unworkable in that state

I don't like the idea of *basing* on anyone else, though. I like the idea of basing on pure Debian and adding Mint tools. I know how it needs to get done, but it takes manpower. As I said above, manpower we don't have, at this moment

@cmost - I don't really have any inclination of spinning my own distro, or of tinkering to that level just for giggles. I only had this specific project in mind, so I doubt I'd be collaborating in any indy spin However, if you needed advice, help, etc... I'd be more than happy to lend you a hand when I can, as I said

[cmost]

As I intimated before, the Mint crowd simply aren't the type to customize and re-spin the distro using an entirely different base. There are a few talented community members who have been able to build Community Editions of Linux Mint for XFCE, KDE and Enlightenment, but they're still using the original Mint base; the Window Manager is only one of many layers that make up a distribution. Obviously it will take a lot more effort, hard work and dedication (and raw talent) to build a Community Edition of Linux Mint based upon an entirely different code base (i.e., Debian.) Furthermore, it's clear that the preference is for Debian Testing. As I said before (and it bears repeating,) it is a paradox to try to stabilize Debian Testing. It simply won't happen. Breakage will occur and unless a system is in place to deal with that then users will find themselves with problems. One only needs to head over to the forums of a distribution based on Debian Testing and read a few posts regarding what happened when a user did a apt-get upgrade to see what I mean. Personally, I would like to see Mint based on Debian Stable with a separate Mint repository to contain updated packages. As you rightly pointed out, 'Lenny' will soon be Debian Stable and its libraries and other packages will be at sufficient versions for Mint tools to work. Since I haven't been bombarded with PM's asking to be involved with creating Debian Mint I'm assuming interest by the community is limited to people wanting the finished product; not those who are willing to help build and test the distribution. We can't have a community edition without the community. I can see why Zonas abandoned this effort. It's too much of a job to go it alone.

[sundayrefugee]

Ideally, MintUpdate would be there to prevent breakage. Users wouldn't be dropping into init 3 and running aptitude dist-upgrade, reading the bug count, and taking their chances. MintUpdate would let them know when it was safe to download a new snapshot That's the whole, entire point of a Mint Debian, and a point I think you're missing - to keep that borkage away from, and transparent to, the end user

But then again, you have to have people behind the scenes breaking things, aptituding bug counts, keeping track of mailing lists, dist-upgrading to see what breaks, etc... Again, that's alot of ongoing maintainance work

You sound pretty upset, though. Hey, we offered ourselves up, in the finest Debian fashion. We were under *no* illusions of the difficulty of the task, and knew it would take a community effort. We knew what had to be done, and tried to see if there was sufficient community help to accomplish it. It turns out there isn't, at the moment.

Don't take it so bad There's still our beloved Mint Boo's KDE is just around the corner. As Mint continues to grow, maybe at some point in the future there will be renewed interest in it. Maybe not.

In the meantime, why not take this time to learn? Good around a bit in virtualbox. Maybe try somewere easy, like remastersys or mklive? Hack a bit with installgui and anaconda. These kind of experiences can be invaluable. Try to do something off-the wall, like turn Mepis into XFCE or Sidux into GNOME. You can't, really, but what you learn is really valuable, and it's kinda fun to boot You'd be surprised what that can teach you. You'll have to learn about unpinning, repinnin, and changing the priority of supplementary repos in preferences files, how to get libs to work with each other, how to aptitude down 15 layers in a tree to get a library to be compatible without changing repos or compiling, how to tag, when to recompile, etc... Then, you're armed with what you lack when the time comes.

[cmost]

You sound pretty upset, though. Hey, we offered ourselves up, in the finest Debian fashion. We were under *no* illusions of the difficulty of the task, and knew it would take a community effort. We knew what had to be done, and tried to see if there was sufficient community help to accomplish it. It turns out there isn't, at the moment.

Don't take it so bad There's still our beloved Mint Boo's KDE is just around the corner. As Mint continues to grow, maybe at some point in the future there will be renewed interest in it. Maybe not.

In the meantime, why not take this time to learn? Good around a bit in virtualbox. Maybe try somewere easy, like remastersys or mklive? Hack a bit with installgui and anaconda. These kind of experiences can be invaluable. Try to do something off-the wall, like turn Mepis into XFCE or Sidux into GNOME. You can't, really, but what you learn is really valuable, and it's kinda fun to boot You'd be surprised what that can teach you. You'll have to learn about unpinning, repinnin, and changing the priority of supplementary repos in preferences files, how to get libs to work with each other, how to aptitude down 15 layers in a tree to get a library to be compatible without changing repos or compiling, how to tag, when to recompile, etc... Then, you're armed with what you lack when the time comes.

I'm not upset in the least about Debian Mint's failure to launch. In addition, I already tinker around with a myriad of operating systems, including Sidux, Sabayon, Project Indiana, every variant of Windows, FreeBSD, several *buntus, and last but not least Linux Mint. I use Mint on my primary desktop because it's simply a joy to use. I probably wouldn't use it otherwise. I'm a fairly experienced Linux user. I'm presently toying with the idea of creating a custom Debian Testing CD from Morphix. I see how that goes.