Today, Ian Murdock described his recent work on APT to the Debian community. This announcement has far-ranging implications for the future of Fedora and Debian projects. Ars Technica has the details.
Today, Ian Murdock described his recent work on APT to the Debian community. This announcement has far-ranging implications for the future of Fedora and Debian projects. Ars Technica has the details.
Imagine the current sid (once gone stable) with an Anaconda install.
Then again we can already shave many minutes using Knoppix HD install. I think installing from a live-cd is the way that big OEM’s install systems now using WinPE and that is the direction longhorn is going.
“It is our hope that a distribution-independent Anaconda and
a distribution-independent APT (plus, eventually, a distribution-
independent configuration framework) will, along with a
stronger LSB, help unify further the various Linux distributions.”
This is AMAZING!!! This would be a gigantic step for Linux mass adoption.
Excelent work Ian!!!
It’s a great start to a new world… a new world wiht out RPMs.
but for the installer… I kind of liked the text-based debian installer. I ment I could install debian to a machine that I had no idea what the graphics card was and it wouldn’t lock up while configuring X.
The this that was missing from the debian install was lspci. The number of times I’ve had to use a knoppix CD or a resuce CD just to run lspci is not funny.
Dispite the great work, I wonder how long it’ll take debian to get off it’s lazy butt to use dpkg-anaconda…
I think its more likely that the Debian APT tools and package developer community will gradually move to Fedora with RPMs than Red Hat will push Fedora to move to DEBs or even a mix of RPMs and DEBs.
In any case, I don’t really care if its a .deb or .rpm as long as the dependency tree is all in place. I just want apt-get to work.
Having a unified distribution would be awesome.
Unless I’m misunderstanding everybody, the idea of unifying fedora/redhat/debian is quite possible the WORST idea ever.
Everybody needs to stop complaining about there being too many distros, too many web browsers, and that KDE and Gnome need to merge. No way.
First of all, if RedHat was able to merge Debian “under” it or buyout mandrake, everybody would be screaming “MONOPOLY!!!”, “MICROSOFT V2.0”, or “EVIL CAPITALISM”. We don’t want one distro. We want competition. Part of why we “like” linux is because it gives us a choice. The more distro’s (to a certain realistic degree) the better.
However, what would be cool (and IMHO is unacceptable that we do not have) is better cross distro compatibility. I believe this is what Ian was trying to point out, so I’m not picking on him. Compatibility is hard enough; unification is impossible and useless. A total unified Linux will in now way increase it’s adoption.
I’m not sure I agree with that.
There are a lot of developing nations in the world where people accept payment in US$ but they are not going to suddently change their entire currency to US$.
It’s like this with .deb and .rpm
You need to re-read the article or just plain read it. APT-GET is being modified to allow it to work with RedHat RPM’s and Debian is getting the GUI installer most people would like it to have for installs.
“There are a lot of developing nations in the world where people accept payment in US$ but they are not going to suddently change their entire currency to US$.”
Exactly. Red Hat will never use debs and Debian will never use rpms just because apt-get can manipulate either. A route to unification would be the adoption of one package format (RPM) and an army of developers to prepare them (Debian->Fedora) for a user community to easily add them to their systems (apt-get).
There is a couple reasons that they probably wouldn’t merge withough significan culture change. Debian won’t switch to RPM’s because .debs are technically “better” afaik, because of better dependancy resolution(I could be wrong here), and the whole debian infrastructure. Red Hat doens’t really want APT because then they couldn’t charge for updates(at least in their pay-for distros) like up-to date or whatever that’s called. Red Hat won’t switch to debs also because they have significan time and mindshare invested in RPM’s.
HOwever, having a common install program and meta-packageing tool would be really sweet. I wonder where autopackage is now?
Is there perhaps a package where the dependecies are described in XML?
We could then use something like XLFST and an XLS-FO to transform this package to either an .deb or a .rpm. It could possibly even be done on the fly.
regards,
Chris
make that XSLT
I don’t think unification is stupid, I think duplication is stupid. Where did it say they were going to join Redhat with Debian? I personally get dispondent when I see an RPM for one distro but can’t find an rpm for the distro I’m using. Its been stated many times the distro’s don’t want to buy each other out. Why do people say Redhat would buy out Mandrake? Redhat runs under USA software laws that aren’t as fair as France laws. So in short by Redhat buying Mandrake, they know they would be asking for trouble in the USA.
I am happy that there is a large choice of
distro’s to choose from but GNU/Linux packaging
is too fragmented and not polished enough
across it’s own spectrum.
GNU/Linux needs a robust and common packaging system
with tar files as an option. we need this to stay strong
and be competitive when fighting M$ and others.
I like to compile stuff from source and tweak it.
However, that willl never gel with corporate
end users. qaulity/simple/smooth/Common packaging
is what will solve a lot of problems , silence critics
and allow the GNU/Linux community be more productive as they
focus on making the big picture happen.
It is more important than most people think!
i think this is a great idea…
Ian sure has a good idea, (sounds good on paper) i will just have to wait till it is implemented in a released distro and give it a spin to see for myself if it is really as good as it sounds…
Installation of packages is exactly what Linux is sooo bad at and Apple and Microsoft are so good at. Sure they’re not perfect but they are far better than Linux. This is perhaps a primary reason for me looking so keenly toward moving to Apple as my primary platform.
The best thing I have heard in a long, long time. Here’s to hoping that it materializes.
Great news, indeed.
It isn’t the infamous installer that keeps me from using Debian on my desktop. It’s the twisty turning maze a user faces trying to build a modern Debian desktop.
Perhaps it is time to consider an different or a supplementive alternative to the stable/testing/unstable paradigm.
Consider: The stable branch is, in fact, very stable. But, from a desktop pespective, it is hopelessly outdated and essentially useless. A user may update to testing or unstable, but that sacrifices stability for — even in the case of unstable — software that is often a few iterations behind existing, widely used, release versions.
A key point, I think, is this: While other distributions build and test a complete mix of current releases, a Debian “stable” user is left to attempt to build and test their own mix of current releases from an array of backports or to migrate to testing or unstable, platforms that — unlike other distributions — lack Debian’s stamp of approval for everyday use.
From the point of view of people responsible for keeping servers alive, Debian’s stable/testing/unstable arrangement makes much good sense. But not from a desktop perspective. Perhaps relabeling “stable” as the “server branch” and opening a new “desktop” branch is worth considering. The testing and unstable branches would remain as is, but software would migrate at different rates to “server” or “desktop”.
It isn’t the infamous installer that keeps me from using Debian on my desktop. It’s the twisty turning maze a user faces trying to build a modern Debian desktop.
Perhaps it is time to consider an different or a supplementive alternative to the stable/testing/unstable paradigm.
Consider: The stable branch is, in fact, very stable. But, from a desktop pespective, it is hopelessly outdated and essentially useless. A user may update to testing or unstable, but that sacrifices stability for — even in the case of unstable — software that is often a few iterations behind existing, widely used, release versions.
A key point, I think, is this: While other distributions build and test a complete mix of current releases, a Debian “stable” user is left to attempt to build and test their own mix of current releases from an array of backports or to migrate to testing or unstable, platforms that — unlike other distributions — lack Debian’s stamp of approval for everyday use.
For me, it just isn’t worth the hassle. If I use Debian, I want to be able to point my apt sources at a few reliable Debian servers and get regular updates that don’t break anything. Jumping into a mix of backports and branches eliminates the simplicity that is Debian’s strong suit.
From the point of view of people responsible for keeping servers alive, Debian’s stable/testing/unstable arrangement makes much good sense. But not from a desktop perspective.
Perhaps relabeling “stable” as the “server branch” and opening a new “desktop” branch is worth considering. The testing and unstable branches would remain as is, but software would migrate at different rates to “server” or “desktop”.
“Dispite the great work, I wonder how long it’ll take debian to get off it’s lazy butt to use dpkg-anaconda…”
The answer is: forever. Anaconda will almost certainly *never* be the official Debian installer. One of the things Debian prides itself on is consistency throughout all its ports, or at least insofar as this is possible. Thus, to be eligible, a given installer would have to have been ported to work on all 11 supported Linux architectures and also be able to support the non-Linux ports, e.g. Debian HURD and Debian *BSD. Perhaps Progeny have extended Anaconda sufficiently for it to be able to handle all of this (there isn’t enough detail to know for sure), but I somehow doubt it.
In any case, a new Debian installer that’s intended to be capable of servicing the entire distribution in this way has been in the works for some time and is now reaching maturity. Progress reports and testing images can be downloaded from http://www.debian.org/devel/debian-installer/getting_started , or alternatively just install Skolelinux (http://developer.skolelinux.no/) which is using the new installer.
Of course, if anyone wants to make *unofficial* install images using Anaconda, nothing’s stopping them. This would actually be pretty useful as a stopgap measure until the next Debian release comes out, but that’s about it.
As far as Progeny’s stated goal of “a distribution-independent Anaconda [to] help unify further the various Linux distributions” (http://lists.debian.org/debian-devel/2003/debian-devel-200310/msg01…) goes… I’m not quite sure how the porting of Anaconda to a distro that’s infinitesimally unlikely to ever adopt it is supposed to be a great first step towards achieving this ambitious aim, but there we are.
“We are also working with various parties to add/merge RPM support into the mainline APT, to allow Debian- and RPM-based distributions to be managed using a single APT codebase, and possibly even to allow Debian and RPM packages to coexist side by side. This work also aims to merge our various APT extensions (e.g., support for authenticated APT repos) into the mainline APT.”
These enhancements sound marvellous, but I’ll save my praise until these things turn into actual achievements rather than just aims for the future.
“Perhaps relabeling “stable” as the “server branch” and opening a new “desktop” branch is worth considering.”
Alternatively, and much more feasibly, a new subproject could be created to tailor Debian for the needs of the desktop. Among (many!) other things, said subproject could mix and match freely from the Stable, Testing and Unstable repositories and support the end result, relieving the end user of the hassles you describe.
And in fact, this has already been done. Check out http://www.debian.org/devel/debian-desktop and take a look, or better yet, join in the fun.
Oh, and as an aside: for certain desktops (most notably workstations in corporate environments) Debian Stable is the ideal choice because of its stability, security and ease of maintenance (the *admins* are going to be the ones setting up and maintaining said systems, so these factors are of critical importance in such a setting.)
I still don’t understand what the jubilation is all about. So debian has a new installer. Big deal!!!
I’m probably falling to flamebait, but i would like to clarify the muddled concept found in this statement:
>>Installation of packages is exactly what Linux is sooo bad at and Apple and Microsoft are so good at. Sure they’re not perfect but they are far better than Linux.<<
Concerning third-party-software, in GNU/Linux terms this is closed source software you have to buy, you may have a point. Browsing to the setup source and depending on your graphical frontend single or double clicking it, entering the root PW and wait is not very comfortable. But at least Debian’s apt (and that’s what we are talking about) does this with one command and most times without further user interaction. Can’t see where the MS Windows way is better. To the contrary, MS Windows user i know complain about the many “Next” buttons to press in an Install Shield procedure.
Whatever…
Considering that this thread evolves around an adaption of apt it is rather astounding to read this. Already now apt is a far superior way to install software than anything Microsoft Windows or Apple OS X has to offer. Type one command (e. g. apt-get install gnome) and apt fetches the software and installs it – often without any need for user interaction. Most MS Windows users i know are annoyed by the Install Shield procedure and comment very favorable on apt.
First of all, I would like to see “unstable” renamed to something like “actively being updated”, preferably in a single word. Debian unstable is not at all unstable in the sense of system or program crashes – it is just that things can noticably change when the system is updated. As a caveat, things do break periodically on updates, usually trivially, sometimes in a major way (two major examples come to mind – one involved a problem with PAM that prevented logins and required booting into single-user mode to fix, another was a typo for the C++ lib that prevented all C++ programs from running). The functionalilty of Debian unstable as a desktop is certainly there – this is posted from KDE 3.1.4 with a 2.6.0-test7 kernel, GCC 3.3.2, etc. It does require some knowledge to fix occasional problems. For corporate use, it would be possible to take a snapshot of unstable, test it, roll it out, and carefully control any updates – probably more work than most places would want, but certainly possible. Anyway, my point is that Debian can be a great desktop.
Regarding the installer – I have no experience with Anaconda, but the single greatest focus for the next Debian release has been a new installer. AFAIK, it is essentially finished for x86 but not for all of the other architectures. I strongly doubt that Debian will abandon all of this work to adopt Red Hat’s installer.
Debian “Stable” is not concerned with the stability of the software itself (that’s what testing and unstable do, among other things) but with the stability (think consistency) of the procedures and software over several different platforms.
Unlike Red Hat or Yellow Dog Debian targets 10 to 16 Platforms (depending on how you count) and on each of those Debian stable behaves the same. That is the stability that is meant by stable – not non-crashing programs – and i guess that you realize that this takes a tremendous amount of time to check and counter-check.
This is something I`d hoped for- an easy installer for debian, bring more users to debian and their wounderful apt-get. And another point that has been mentioned here- one hell of a stable “stable” server version and one destop version.
Desktop version using the RedHat installer, server using the “new debian” installer-this due to the fact that RedHat installer is not so great on different platforms ( so I`ve read somewere).
“That is the stability that is meant by stable – not non-crashing programs”
Actually, it means both. Consistency is certainly a key aspect (e.g. no updates other than security fixes) but the release process is also bug-driven (see http://bugs.debian.org/release-critical/); no packages can be released whilst they or any of their dependencies have RC bugs filed against them. Whilst bugs of a lower priority don’t hold up the release, they are either forwarded upstream or fixed by the maintainer or anyone else sufficiently motivated to provide a patch or a solution.
Oh, and a program crashing would almost certainly be classified as a release-critical bug, so have no fears on that score – any package suffering from such a problem won’t be included in the release.
Syntaxis: Thanks for the pointer to that desktop project. But Here’s one thing I don’t get: Unless I’m misreading the page, why is XFree86 still called “experimental”? The rest of the world has been using it quite happily for a rather long time. Does Debian know something no one else knows? Frankly, that’s what befuddles me about Debian. I’ll have to disagree, too, about the appropriateness of stable for enterprise desktops. It may reduce admin workload, but everyone I know would run screaminng in horror from stable on their office desktop. They’d refuse to use it.
Somewhere: I understand that Debian supports a variety of architectures. That’s admirable. But, the implication of what you’re saying is that progress in the x86 arena — where most of the world lives — is artificially constrained to assure consistency across all platforms.
With the creation of Fedora, RedHat has now clearly seperated its experimental/testing branch from its stable commerical branch. Although somewhat akin to Debian’s stable, unstable, and experimental, it is more more akin to the analogy that Fedora becomes to RedHat, what Debian is to Libranet, Lindows and Xandros. Moreover, Fedora now brings a number of hitherto quasi-proprietory RedHat tools into the wide open, esspecially the excellent Anaconda installer. This is esspecially good for a company like Progeny which specializes in customizing Linux for their clientel.
So Ian Murdock, in his genius, thought rather then work from a base of RedHat or Debian, as so many variant distros do, they could now work from a unified base of Fedora/Debian, if three issues were addressed. 1) a common package manger, 2) a common installer and 3) a common configuration framework.
On the first account developemnt had already began. With the advent of alien and then apt-rpm, it’s but a stones throw to roll .rpm management, along with .deb, into a unified Apt. And suddenly the whole package management debate is a hairsbreath from resolution (Hallelujah!).
On the second, Fedora gives us Anaconda which is about as close to a perfect installer as one could ask. With a little alteration it works just as well for Debian as it does for RedHat. Now Debian’s antiquated installer can finally drop dead. (Another hallejuah!)
And lo, we are but one step away from a unified Fedora/Debian. Interestingly it wasn’t by any direct intent to unify the two per-se, but rather only to expedite Progeny’s work. I think this actually says something about the whole Open Source and GNU/Linux movement in general. For one, there has been much concern the GNU/Linux would fail for the same reasons UNIX did. But now it is clear, the Open Source community, which UNIX lacked, is exactly what will prevent this from happening. Secondly, and more importantly, it shows the a standards base for GNU/Linux is developing naturally out the needs of its users. Although LSB tries to artifically create a standards base from the top down, and succeeds to some degree, there is also the opposite natural tendency for standards to develop from the bottom up. Between the two, GNU/Linux is, slowly but surely, developing a real standards base.
But the fullfilment of this improvement depends on one important factor: adoption. There will no doubt be a backlash against Progeny’s work. But I must implore all parties involved to seriously consider what it means to our community: the birth of a true standard.
This means, among other things, that those involved with LSB should follow the natural flow and embrace the unified Apt and allow .debs into the specification. From there they should concentrate on a common configuration framework. For Debian developers, this means we must graciously ask them to move to Anacanda. I know the developers who have worked so hard on the new Debian installer will bulk at this, but it is sometimes a hard fact of life that we must, figuratively speaking, kill our babies. And Fedora developers should embrace the new unified Apt (as much as I like the word Yum). Together, and with the adoption of Discover, which has already become somewhat of a defacto-standard for hardware detection, GNU/Linux will be “better, faster, stronger” then ever before. Which is something we all want.
If you ask me, it’s a great day in the world of GNU/Linux!
XFree86 is not referred to as experimental. The unstable XFree86 4.3.0 packages are currently in the experimental branch since they’re still being tested. Clearly it’s taking a while for 4.3 to get in to sid, but if you really need to use 4.3.0 instead of 4.2.1, there are plenty of unofficial package repositories you can use (http://www.apt-get.org), and the same goes for most popular software held back by testing/bugs etc. in testing/unstable.
I don’t understand why everyone thinks .RPMS are any worse than .DEBS. I think it’s a misunderstanding that .DEBS have some sort of magical dependency checking that .RPM’s don’t. APT does the very same thing for DEBS that it is doing for RPMS. Without APT debian would have the same dependency hell that people claim RedHat has. This installer deal is good news!
i agree, i been a rpm user for a long time without problems, occasionally i run in to a dependency problem, all i do is save my rpm in a temp directory go to http://www.rpmfind.net download whatever rpm is required to complete the install and use a root shell in that directory and run rpm -iUv *rpm and the problem is solved…
Man, your post was hilarious. Nice one! 😀
Note: I’m assuming that you’re not being serious (lol – if you are, you should seek help) but I responded as if you were, just in case. 😀
“Although somewhat akin to Debian’s stable, unstable, and experimental”
First, you missed out the “Testing” repository, but more importantly, how is it even remotely similar? Having a separate repository for development is standard practise, and is nothing special. To give just two examples: before Fedora, RedHat had RawHide, and Mandrake has the Cooker.
“Fedora becomes to RedHat, what Debian is to Libranet, Lindows and Xandros.”
Perhaps, in time. Debian’s commercial offshoots dwindle to insignificance when compared to the projcet itself, whereas with Fedora and RedHat the relationship is (for now) the other way around. In addition, Debian is owned and controlled by no-one save its development community, whereas Fedora is effectively owned by RedHat and they have simply opened the development process to outsiders without relinquishing their control. People may argue over whether or not this is an important distinction, but it nonetheless is a clear difference between the two distributions.
“Fedora now brings a number of hitherto quasi-proprietory RedHat tools into the wide open, esspecially the excellent Anaconda installer”
No. Every component of their distribution was released under an Open Source license prior to the creation of Fedora. No change here.
“1) a common package manger”
They haven’t come anywhere close to achieving this yet, they merely stated they are working on it. Wait ’til its done before singing their praises.
“2) a common installer”
As I stated earlier, Debian, the very distribution they ported the installer over to, cannot use Anaconda as their official installer because it doesn’t support all the ports/architectures. Progeny have also left it a bit late – the new Debian installer is almost finished.
“3) a common configuration framework”
Again, I’m hopeful, but I’ll believe it only when I see it. In addition, they said they hope to achieve this “eventually”, so I wouldn’t hold your breath at seeing this any time soon.
“Fedora gives us Anaconda which is about as close to a perfect installer as one could ask. With a little alteration it works just as well for Debian as it does for RedHat.”
Lol – once again, shame about all those ports/architectures it doesn’t support.
“But I must implore all parties involved to seriously consider what it means to our community: the birth of a true standard.”
There’s no conspiracy here. Make it so that it clearly *is* far superior to any existing solution, and people/distributions will migrate of their own volition. As it stands, they won’t, because it’s not.
“those involved with LSB should follow the natural flow and embrace the unified Apt and allow .debs into the specification.”
The LSB already allows this. To be compliant, a distribution merely has to support RPMs. If they choose to support any other formats on top of that, that’s their perogative.
“For Debian developers, this means we must graciously ask them to move to Anacanda.”
Lol – please make sure to do it on one of Debian’s public mailing lists. The response should be most entertaining, to say the least. Make sure to wear your flameproof suit as you do so, naturally. 😉
“Together, and with the adoption of Discover, which has already become somewhat of a defacto-standard for hardware detection”
No, it hasn’t. RedHat uses Kudzu, for example.
@ Aitvo:
The RPM format doesn’t support “suggests” or “recommends” package tags, for one thing. I certainly miss those when I’m using an RPM-based distribution.
@ Cheapskate:
“occasionally i run in to a dependency problem, all i do is save my rpm in a temp directory go to http://www.rpmfind.net download whatever rpm is required to complete the install and use a root shell in that directory and run rpm -iUv *rpm and the problem is solved…”
The trouble with that is there’s no assurance of the quality of the packaging, or even as to the contents. There’s *no* QA whatsoever. Any Tom, Dick or Harry might have packaged that RPM you’re installing. It might install to the wrong place, it might not even work at the end of it, and they might have trojaned it while they’re at it, too. Thankfully Fedora will ensure that all packages within the repository undergo at least a basic level of QA, though still nothing compared to the Debian packages atm.
“about the appropriateness of stable for enterprise desktops. It may reduce admin workload, but everyone I know would run screaminng in horror from stable on their office desktop. They’d refuse to use it.”
Provided that they are able to provide the staff with the necessary functionality to get their typing/email/web browsing done, it’s up to the IT department to decide what distribution the workstations will run. And the IT support staff will obviously be inclined to opt for something that makes *their* jobs so much easier – that’s just human nature. Even if they really must opt for backports to get some piece of extra functionality, or eyecandy (which is sadly how most end-users seem to judge software quality) it’s still a joy maintenance-wise compared to other distributions. Lol – but fair enough, I accept that not everyone will see it that way. 🙂
“But, the implication of what you’re saying is that progress in the x86 arena — where most of the world lives — is artificially constrained to assure consistency across all platforms.”
No, not at all. It’s simply that the Debian project cannot afford to have a different installer for every port. Not only would it be monstrously inefficient due to duplication of effort (and thus a waste of developer time), it would mean that the users would have to master a different install procedure for each port.
And who’s “constraining” you? AIUI, you’ll be able to apt-get Progeny’s creation from the archive, whereupon it’ll generate install images for you automatically. After that, you can freely redistribute them if you so wish. In addition, although it was not mentioned in the announcement, I suspect that Progeny will have some install images available on their site, as they did for their old installer, PGI.
Even if this doesn’t bring Fedora and Debian closer together, Debian will now contain RPM compatibility via apt instead of just alien. It will also include another installer for us to allow our lesser knowledgable freinds install our favorate distro. I see no down side to this, Ian is a god!
Best case scenario, RedHat see’s how useless Fedora is, releases everything to Debian, Debian becomes the central place for all things Open Source. OK, I think I went a little far with that one, but this does seem like what Progency is leading to…
I don’t think this announcement will really effect anything to much though, other then RPM APT repositories being accessable to people running Debian that don’t know any better then to use RPM’s…
/me would like to see Debian being the central place for all community based developement in the open source community!
God has listened to my prayers!! Debian and Fedora united! Yess!
This is *awesome* news. The new installer is definitely great stuff. But the best news, in my opinion, is Debian and Fedora going for a common repository! I think having so many different packages is a BAD thing for Linux. It would make everything more easy if we would all use a single package.
I would love to see Fedora using .DEBs, but i think that’ll be hard to happen… this is what i think they should do: drop DEB, drop RPM, create a new one (DFPM – Debian and Fedora Package Managment), or use autopackage!
Ian, you rock!
Victor.
It is not stupid since the packege manager should be
unifide not the whole distro. You could install a *.deb on
fedora and a *.rpm on debian. Then we could stick to one
of the two formats and live with out dep hell. The system
layout should be the same for all Distro’s
Eg.
/etc for config files
/boot vmlinuz-2.x.xx and other files for booting.
And so on.
Yes Debian will still be solid
Yes suse will still have yast2 and outh tools
Yes Gentoo will still be source based.
And yes Fedora will still have that clean bluecurve.
Aaron
Syntaxis, you must live on a different planet than I do. There’s not a single program in an unadulterated Woody install that a typical corporate user would know how to use, or would wnat to use. Even if they did know how to use, say, vi, what good would that do them? Besides, XFree on Woody looks like crap. People who have to spend all day staring at their screen would rebel.
And, I’ve never seen an outfit that allowed the IT staff to decide what software users run. The CEO is a user. You wanna tell the CEO he/she can’t use Office and Outlook? For that matter, does Lotus Notes run on Woody??
Geez, would you expect the CEO and the board to buy a second-rate air conditioning system just because it made life easier for the maintenance crew?
Meanwhile, it isn’t sad that users judge sofware quality based on “eye candy”. First of all, ease of use and clarity of display is not eye candy. Second, users tend to equate inattention to display quality with inattention to performance quality. They are not necessarily wrong.
Finally, I appreciate that a volunteer effort like Debian lacks the resources to create and support different installers for different platorms. Fair enough. And, yes. platform-specific installers might confuse people supporting multiple platforms. But, honestly, how many people really do?
As for me, I like the basic premise of Debian, and I very much wish I could use it, I’d also like to use the applications everyone else is using in the year 2003, and I’d like to use them on a desktop that doesn’t remind me of 1995. Sadly, I can’t seem to do that in Debian unless I venture out on a problemmatic path.
“www.rpmfind.net is a legitimate and well known website containing rpms for several distros”
If you say so. Now just point me to which area of the site lists the minimum quality requirements that a package must meet in order to be indexed there, and the rigorous and consistently-applied QA process that is used to check that the packages do in fact meet these requirements. I don’t think there *are* any such requirements or quality control, but feel free to prove me wrong.
“There’s not a single program in an unadulterated Woody install that a typical corporate user would know how to use, or would wnat to use. Even if they did know how to use, say, vi, what good would that do them?”
Well, at least you’ve now dropped the pretense that you’re doing anything other than blindly trolling. You’d be hard pressed to find an editor less intuitive for the uninitiated than vi.
“And, I’ve never seen an outfit that allowed the IT staff to decide what software users run.”
Then you need to open your eyes. To give an extreme example, many corporations have IT policies that mandate the use of a given piece of software as standard throughout the organization, e.g. Outlook and Microsoft Office, because they believe that doing so will help cut down on support costs. If a user’s personal preference is for a different piece of software, that’s just too bad.
“You wanna tell the CEO he/she can’t use Office and Outlook? For that matter, does Lotus Notes run on Woody??”
We were talking about Debian Stable compared to other distributions, not compared to Windows. However: Crossover Office runs just fine on Woody, as do OpenOffice and StarOffice. There’s also a wide variety of mail clients available to choose from. I think KMail and Evolution (to pick just two of the best known) are pretty intuitive, and Evolution’s UI especially bears a marked resemblance to that of Outlook.
“Geez, would you expect the CEO and the board to buy a second-rate air conditioning system just because it made life easier for the maintenance crew?”
“Second-rate” is the wrong word to use. To buy a better air conditioner than is necessary to adequately air-condition the building would be a waste of money which could be better allocated to where it is actually needed. Also, I would have thought the CEO and the board would have better things to do than decide which air conditining unit to buy – they’d just let the appropriate department/personnel handle it. Oh, and if increased ease of maintenance requires fewer/less experienced support staff on the payroll (and it does), yes, I suspect they’d be all for it.
“Meanwhile, it isn’t sad that users judge sofware quality based on “eye candy”.”
Yes, it is. It’s an entirely superficial criterion. The most important thing by far is that the program is stable and provides the required functionality to get the job done. It’s just a tool. Eye candy is a nice bonus but far from essential, and a pretty program can still be a hunk of junk. Insert “judging a book by its cover” analogy here…
“First of all, ease of use and clarity of display is not eye candy.”
I don’t think such things as window resizing effects, splash screens, and widget translucency are that essential for either ease of use or clarity of display. Font anti-aliasing is debatable, but in any case the versions of KDE and GNOME in Woody do have support for this.
“And, yes. platform-specific installers might confuse people supporting multiple platforms. But, honestly, how many people really do?”
A hell of a lot of people. Apple hardware’s pretty darn common these days. I also see a lot of SPARCs kicking around, as well as a few Alphas. Besides, since Debian’s focus on consistency makes supporting multiple platforms easy, it would make sense that people who need/want this would gravitate to the distribution.
“I’d also like to use the applications everyone else is using in the year 2003”
Then either 1) get packages from Unstable 2) help out with the Desktop subproject 3) make use of the wide variety of backports from http://www.apt-get.org or 4) just bite the bullet and use what’s in the Stable package tree. Most people would find the functionality of a one-year-old email client quite sufficient for their needs, for instance. Heck, the majority of people are still using the Outlook Express that came with Windows 98.
“and I’d like to use them on a desktop that doesn’t remind me of 1995.”
Woody was released on the 19th of July, 2002, not 1995. So it’s really nowhere near as antiquated as you’re making it out to be, especially in a corporate setting where the software typically has a three year upgrade cycle, if not longer.
You say you want Debian to be the center for the open source community?
Debian is an organization where, in order to get in, you have to pass a ideology exam (check it out, requisites to become a maintainer).
Making Debian the center of anything is insane.
From http://www.debian.org/devel/join/nm-step3, under “A. Philosophy”:
“Note: The term “is expected to agree with the Debian philosophy” only applies to the work the applicant will do for Debian, and the ideals they will use when doing Debian work. These principles and guidelines are established expressly for the Debian Project.”
You’re simply being asked to pay lip service to the Free Software Guidelines and Social Contract and show understanding of what they mean and represent. Provided you don’t sound like you want to subvert the project from within or attempt to sneak Non-Free software into main, you’re pretty much in the clear. This doesn’t sound too onerous to me.
It’s simple. If the work you intend to do related to open source software disagrees with Debian philosophy, you are expected to do it outside Debian.
Which is not onerous… unless Debian is “the central place for all open source development in the community”.
Simply put, a community which refuses acceptance of those who disagree with its philosophy is a closed community, not an open one.
So, I hope Debian stays where it is, in the periphery, and never becomes “the central place” for anything.
That way, those who agree with their philosophy can go plain in that restrictive sandbox, while the rest of us are free to roam.
“If the work you intend to do related to open source software disagrees with Debian philosophy”
I suppose this might hypothetically be possible, but… Can you please come up with a plausible example of how promising to follow the spirit of the Free Software Guidelines and Social contract in one’s official capacity as a Debian developer would create a real conflict of interest with regards to Open Source software development? I’m finding it hard to think of one, but perhaps someone else can.
Of course, Debian’s Social Contract is with the Free Software (http://www.debian.org/social_contract#guidelines) not Open Source (http://www.opensource.org/docs/definition.php) community anyway, if you want to be pedantic about it. While there’s a lot of overlap, the two organisations do judge licenses by different criteria.
“Simply put, a community which refuses acceptance of those who disagree with its philosophy is a closed community, not an open one.”
Despite its stated goal of creating a 100% Free Software distribution, the project provides a tree for Non-Free software. I’d be hard pressed to think of a greater concession to those that think Non-Free software has its place in the software ecosystem, so I don’t think your point is valid.
“unless Debian is “the central place for all open source development in the community”.”
The person you’re quoting immediately chastised himself for exaggerating (“OK, I think I went a little far with that one), so I don’t think it’s fair to take that literally anyway.
“So, I hope Debian stays where it is, in the periphery”
It’s hardly in the periphery. Whilst it is of course absurd to suggest that Debian *is* the Free Software community, it is undeniably one of the major players within it, and one of the largest embodiments of the community spirit. Above all, it’s a wonderful example of how community self-governance can be effective. It also serves as one enormous testbed (and therefore, development platform) for the packaged software within the archive, thanks to the maintainers’ close liason with the upstream developers. In addition, the fact that there is no comparably successful community-developed distribution speaks volumes in itself.
No trolling here, Syntaxis. I said that there isn’t a single program in a Woody installation that a typical desktop user would know how to use, or want to use. Vi is one example from among many. It’s hard enough to sell people on Linux as a replacement for their Windows desktops. (Especially when the replacement is a clone of Windows and adds no new functionality. Why go to the pain of switching only to end up where you started?) Show an XP user a Wooody desktop and they’ll walk away.
Yeah, I’ve seen lots of organization mandate use of particular programs, but the mandating wasn’t done by the IT department. Any software used on the desktop as part of the daily workflow was mandated by management, i.e. the people the IT folks work for. I’ve also seen organizations in which IT unilaterally replaced a major component without explanation and with inadequate notice. It almost always resulted in a great deal of boardroom angst, especially from managers who saw their staff production dropping and their training costs increasing.
Aren’t the Woody mail apps you mentioned hopelessly outdated? If you actually do convince someone to invest in Linux, why wouldn’t they opt for an up-t0-date distribution?
Ease of use and display clarity have nothing at all to do with gimmickry like windows transparency, splash screens, and the like (although it seems many Linux advocates are confused on that point.) If the quality of the display is lacking, people are turned off. No one wants to watch TV if the picture is fuzzy. No one wants to use a computer if the display is fuzzy. And the display on a stock Woody install, using the ancient XFree86 code that comes with it, is fuzzy.
Then there’s the questions of useability and design. Applications from the Woody era set the example for how not to do it.
Here’s my basic point: You keep telling me that stability is the primary thing because it lowers ease and cost of maintenance. In other words, makes life easier for admins. (While, I argue, making life miserable for users.) But, what is the trade off? You force people to run outmoded software and what do they get in return?
If the only way to crank out a stable Linux distribution is to do it like Debian, that must mean that all the other modern distributions are unstable. But, they aren’t. So, what’s the deal?
On a more serious note, I beleive that one great thing that should come of this is for APT to maintain .deb and .rpm databases for all packages installed. I think this should be defined in the LSB. Its stupid to restrict everyone that is LSB complient to just one package format, but its not stupid to think that APT can maintain the databases for all package formats.
I think it definatly seems this will happen on Debian systems, if only because the LSB currently states we must support RPM’s, but why not make something like APT mandatory for all distro’s? Makes more sense to me then making the RedHat Package Manager mandatory. Advanced Package Tool seems much more distro non-specific…
Anyway, I have said since Fedora was created that it should work with Debian (I have been on #fedora on freenode and debated this). It just makes sense, and Fedora project and RedHat Projects were merged because “the goal was too simular to maintain seperatly”, well Fedora and Debian are pretty simular also. Only difference being that Debian is controlled by the community, and Fedora is controlled by a distro. If RedHat takes away its control, they are essentially the same.
And to the person that said that its crazy to have tests etc to be able to maintain for a project, you would rather any fool be able to maintain the packages you install? I think not. I know I don’t want any packages on my system that was created by a fool!
Anyway, afaict, none of this has been uploaded to unstable yet. When/if it does, I think this will definatly be a good thing no matter what ends up happening with it.
i started out as a windows user with Win98se on my first OEM computer, then when WinME was released i seen what it was like on my brother’s computer, it would occasionally lockup & freeze and BSOD worse than my Win98se would, so i did not buy WinME, when XP came out there was the uproar about Product Activation so i did not buy XP and bought Redhat-7.1 which was my first taste of Linux, it ran great, no lockups, no BSODs, no need for a anti-virus, the firewall was built in to the kernel, i tryed other distros, Slackware is ok but takes a little more knowledge and work, Debian can be a nitemare to install if you are not carefull, gentoo’s install is more than the avarage computer user has patience for (especially me) well after trying lots of distros, i find myself sticking with my old favorite Redhat now using JAMD Linux http://www.jamd-linux.com which is a clone of Redhat-9 that has been tweaked a little and had some of the bloat trimmed off to make it all fit on a single CDrom, i can not say Linux is going to be the answer for everyone, but i can say it is the answer for LOTS of people out there allready, if you never used it i say give it a spin, or at least dualboot with your existing Windows install so you can “have your cake and eat it too”,
good luck &
HappyTrails :^)
Cheapskate, I don’t dual boot and I don’t run Windows. I’m using FreeBSD at the moment, but I’ve also used Linux since the mid-1990’s: Red Hat, Debian, Slackware, Gentoo, Mandrake, Arch…the lot. Of them, Slackware is my favorite; when I use it, I prefer to do a minimal install and then compile what I want from source.
My issue is not with the Debian install. It’s with Woody’s lack of contemporary applications and with Woody’s use of an old and outdated version of XFree86. Debian has a lot to offer, but I’m not prepared to step back to 1997 to get it.
Woody is lauded for stability, but the trade off is that you’re saddled with old code and, frankly, an ugly desktop. If you want to change it, you need to look at backports, etd., because even the unstable branch isn’t really current. Once you start looking at backports, or installing from source, you’ve lost the two chief benefits of Debian: Woody’s stability and the use of apt.
Other distributions seem to be running quite reliably without sacrificing quality and contemporary applications.
It isn’t unusual for some folks to downplay the display quality issue, casting it as “eye candy”, but that’s just dead wrong. People want quality. Anyone’s first impression of an computer system is what they see on the screen. Expectatons will be lower if the display fails to impress.
Look at it this way: Put an XP, OS X, and Woody machine in front of someone. Load Yahoo in IE on XP, on Safari in OS X, and Netscape on Woody. Ask them which display is the least pleasing.
I love Debian use it every day, but Gentoo is more advance in many way and more developed.
Apt-get is cool but emerge is great and blast.
“Debian has a lot to offer, but I’m not prepared to step back to 1997 to get it.”
Lol – well, the “lot to offer” part is certainly spot on but, once again: Woody was released in *April 2002* – not 1997. But hey, you said 1995 last time, so you’re getting closer.
When someone has to consistently resort to extremes of exaggeration to make any kind of a point, it’s pretty clear that they’re blowing smoke.
“I said that there isn’t a single program in a Woody installation that a typical desktop user would know how to use, or want to use.”
Yes, and that’s a troll. It simply isn’t true.
“Vi is one example from among many.”
And I say again that anyone who uses Vi as an example in an argument about desktop suitability is a troll. There are many more intuitive editors to choose from.
“Especially when the replacement is a clone of Windows and adds no new functionality. Why go to the pain of switching only to end up where you started?”
To no longer be tied into software upgrade cycles, to no longer be locked into a single vendor, for better security, and lower costs of ownership due to ease of maintenance, among other things. If it really is a clone of Windows in functionality as you say, whilst providing all the advantages I just mentioned, it would be insane *not* to migrate, hmmm?
“Yeah, I’ve seen lots of organization mandate use of particular programs, but the mandating wasn’t done by the IT department.”
Don’t play silly word games. The IT department is responsible for the IT side of things. The fact that the department will have a manager in charge is obvious and irrelevant.
“I’ve also seen organizations in which IT unilaterally replaced a major component without explanation and with inadequate notice.”
Good for you. That’s irrelevant however, as it’s got nothing to do with what we were talking about.
“Aren’t the Woody mail apps you mentioned hopelessly outdated?”
No. Again, Woody was released in April 2002. And as I said earlier, Outlook Express is even more outdated, yet the majority of the desktop users in the world still make use of it.
“Ease of use and display clarity have nothing at all to do with gimmickry like windows transparency, splash screens, and the like”
Great, no disagreement here.
“And the display on a stock Woody install, using the ancient XFree86 code that comes with it, is fuzzy.”
Erm. No, it isn’t. I’ve never experienced this problem, so it must be due to your hardware or configuration. Also, the version in Woody is XFree86 4.1, which was the latest upstream release until January 2002. In other words, it’s really not that old.
“Applications from the Woody era set the example for how not to do it.”
Lol! Era? Once again, Woody was released in April 2002, not in 1997 or 1995 as you’ve claimed in this thread. Besides, without specifics, you’re just casting empty aspersions.
“You keep telling me that stability is the primary thing because it lowers ease and cost of maintenance. In other words, makes life easier for admins.”
Yes.
“(While, I argue, making life miserable for users.)”
Lol. Okay. I argue that it doesn’t. The software’s perfectly fine, and the aesthetics are perfectly acceptable as well. A quick Google will bring up lots of screenshots to back me up. Besides, an IT department with less time being eaten up with day-to-day administration will have more resources available for responding to queries and support requests, in other words, the *human* side, so everybody wins.
“If the only way to crank out a stable Linux distribution is to do it like Debian, that must mean that all the other modern distributions are unstable. But, they aren’t.”
Yes, they are. Debian *Unstable* is for the most part as stable as any of the other distributions, let alone the Stable tree. In addition, Debian is the only distro with a feature freeze policy re: its releases (this is what makes administration so painless – no software upgrades to deal with) and it has the largest software archive, the most thorough packaging and QA process, and a large and friendly support community around it thrown in to boot.
Syntaxis, you seem to have an odd emotional attachment to Debian.
This is not difficult:
Woody installs with versions of popular open source software that are several iterations behind other Linux distributions. That’s a fact.
Some of these applications are available from independent developers, outside the official Debian project.. There is no guarantee that Backport A won’t conflict with Backport B. And, obviously, Debians stable tree doesn’t track these backports. That’s a fact.
There are no Debian security updates for backports, as well as testing and unstable. If you want to maintain that famous Debian reliability and security, you have to stay with the Debian stable tree. That’s a fact.
Woody’s selling point is security and reliability. In return, you get older apps and an old display. If you think that’s a fair trade off, fine. But, if someone wants Woody-level stability and reliability and the ability to rely exclusively on official Debian updates and Debian security patches and at the same time run the current version of XFree86 plus Gnome or KDE, Mozilla, OpenOffice.org, etc., they can’t.
If you know about a site that offers instructions in how to upgrade Woody — using official Debian sources — to XFree86 4.3, KDE 3.1.4/Gnome 2.4/xfce 4/, Mozilla 1.5, and OpenOffice 1.1 please let us know.
Now, as for your suppositions about what corporate workers would or would not use on their desktops, please let us know why an average employee would choose Woody if the alternatives included OS X or XP. Not what an T department would choose if they incorrectly thought they’re only imperative was to cut costs, but what the users would choose.
“There are no Debian security updates for backports, as well as testing and unstable. If you want to maintain that famous Debian reliability and security, you have to stay with the Debian stable tree. That’s a fact.”
Yes, and wait for the next version of the software which has the bugfixes
“If you know about a site that offers instructions in how to upgrade Woody — using official Debian sources — to XFree86 4.3, KDE 3.1.4/Gnome 2.4/xfce 4/, Mozilla 1.5, and OpenOffice 1.1 please let us know.”
s/stable/unstable/g in /etc/apt/sources.list and apt-get update; apt-get dist-upgrade
If you want XFree86 4.3, you have to get it from experimental.
I write this with Mozilla 1.5 from an XFree86 4.3 and Gnome 2.4 debian desktop and haven’t ever had a single crash.
Gabriel, I understand that you can pull from unstable, but, once done, you aren’t running the stable branch any more.
While there may be a great deal of anecdotal evidence that “testing” and “unstable” are reliable enough, but, per the Debian site, stable is the “official” release, “testing” is “not completely tested and has no official support from Debian security team”, while unstable “is run by developers and those who like to live on the edge”.
“Syntaxis, you seem to have an odd emotional attachment to Debian.”
Yes, I’m biased. I do like Debian; well spotted. 🙂 But hey, I’m extremely moderate as far as distribution advocates go (I can <gasp!> admit that Debian may not be absolutely perfect in all respects). 😀 I also know a good deal about the distribution due to my interest in it, which makes me a useful participant in discussions about it (or so I like to think).
“Woody’s selling point is security and reliability. In return, you get older apps and an old display. If you think that’s a fair trade off, fine.”
Fair enough; well put. One obviously cannot have a feature freeze policy without falling behind in version numbers compared to the competition. However, your earlier assertions that all the software is so antiquated as to be completely and utterly useless remain bogus. “Older” = 2002, not 1995 as you previously claimed.
“But, if someone wants Woody-level stability and reliability and the ability to rely exclusively on official Debian updates and Debian security patches and at the same time run the current version of XFree86 plus Gnome or KDE, Mozilla, OpenOffice.org, etc., they can’t.”
Sure. I agree with you that the resulting system will be *less* stable and secure than a system installed from pure Woody sources. However:
1) Those who use other popular distributions e.g. RedHat or SuSE, usually install un-official packages from FreshRPMs or similar on a far more regular basis without even thinking twice about it.
2) Many of the backports listed on http://www.apt-get.org are provided by reputable sources such as Debian package maintainers, and some software projects also provide their own packages for Woody – KDE is one such. This mitigates the risk somewhat, and the level of support remains quite high.
3) Additionally, only the backported packages themselves are affected – the rest of your system remains Woody, with all the advantages that entails.
“If you know about a site that offers instructions in how to upgrade Woody — using official Debian sources — to XFree86 4.3, KDE 3.1.4/Gnome 2.4/xfce 4/, Mozilla 1.5, and OpenOffice 1.1 please let us know.”
XFree86 4.3, maintained by a Debian developer: deb http://people.debian.org/~mmagallo/packages/xfree86/$(ARCH)/ ./
KDE 3.1.4: get the debs from http://kde.org/download (duh!)
Mozilla Firebird 1.6.1: deb http://www.backports.org/debian stable all
OpenOffice 1.1, courtesy of the official Debian OpenOffice maintainers: http://linux-debian.de/mirrors.html
XFCE4, courtesy of the official Debian XFCE maintainer: deb http://people.debian.org/~madkiss/xfce4/ ./
The Gnome backport is still at 2.2. Nonetheless, it has a large userbase (requiring several mirrors) and is well supported: deb http://ftp.acc.umu.se/mirror/mirrors.evilgeniuses.org.uk/debian/bac… gnome2.2/
There *is* a Mozilla 1.5 backport, but a) I use Firebird and b) it was by someone I never heard of, so I don’t trust it enough to list it here.
As it stands, all those repositories above, whilst unofficial, are maintained by what I would consider reputable sources. I’ve also used most of them myself at one time or another. I would highly recommend uninstalling the Debian versions of the packages before installing any backports, however, just in case they conflict.
One fair criticism of the backports system is that it could benefit from being rather more centralised. http://www.apt-get.org provides an extremely useful service, but it would be nice if the Debian project officially recognised backports’ existence and provided an equivalent indexing resource somewhere within the debian.org site. It would also provide an opportunity for better QA. Using only packages that come from sources you trust is a reasonable metric to avoid having one’s fingers burnt, but the flip side of the coin is that adoption of otherwise excellent backports may languish simply because those providing them have no official role/recognition within the community.
“please let us know why an average employee would choose Woody if the alternatives included OS X or XP”
I wouldn’t expect the average user to choose Woody. 😀 The primary reason being, I wouldn’t expect him to have even the faintest idea of what it is. However, I *would* expect the average employee to be able to use it to get his work done, once the admin had set up the workstation ready-to-go on his desk. It may be less visually appealing than other alternatives, but it’s certainly nowhere near as ugly or unintuitive as you were making it out to be.
Besides, I thought you were saying that *management* are the ones who make the decisions? If so, I still contend that the choice that offers the necessary functionality (remember: a computer is just a tool) at the lowest initial and ongoing cost will be the clear winner. This means that if the alternatives provide functionality over and above the minimum that is required, whilst being more expensive in terms of initial outlay and/or maintenance, they’ll still lose out.
To sum up: you correctly expounded upon a couple of Debian’s weak points, but I hope that I’ve adequately demonstrated that these shortcomings are a) a worthwhile trade-off and b) nowhere near as problematic or severe as you were initially making them out to be, resorting to excesses of hyperbole as you were. Lol – but, whether I’ve succeeded or not, this debate has been so entertaining that in the end I’m not really bothered either way. 🙂
I’m running 2.4, and things like drawers aren’t working properly. It’s probably a good thing that 2.2 is still the current.
Although I was severly burned for my last post :p I will give some relevent info that I hadn’t seen mentioned here yet: that of the state of the Debian Installer.
I gave the sarge installer a go on an old box I had about. I put the CD in, booted up and was introduced to the common boot: prompt with info on how to proceed. I hit enter. The installer reported the it was installing installer modules. A bit redundent not to mention quick to put files on my drifve (where did they go?) But okay. The next step was a nice text based question about my language (lots to choose from!) then there may have been another simple question, I don’t recall. Next was setting up my network but it failed to detect my network card. great. went through every possibility, none of which worked (i have a siemens card in that box) in the course of trying every possibility i kept getting some odd unrelated message. finally i was told to load a driver from a floppy. uh, what floppy? i don’t even have a floppy drive. And so I was able to get to the 3rd step before failure. Lovely.
I put in my knoppix disk and played frozen bubble instead.
Yea, Ian’s Anacanda idea, that’s a bad one. sic.
I don’t know if anyone else noticed this, but shouldn’t the title here be “Progeny Debian Brings Red Hat and Debian Closer Together” after all thats the name of the project “Progeny Debian”, not Progeny Linux.
Anyway, not that it really matters, but I would like to see it changed. Is like saying RedHat BSD or something…
It is kinda a big deal though, on account of Progeny meaning “comming from” and all, Progeny is Ian’s developement of HIS distro as a company… Debian still gets all the development aspects of Progeny, and Ian still gets to make money…