Weekly Edition
Return to the Distributions page
| |||||||||||
A new UserLinux draft manifesto
Bruce Perens has posted a
new UserLinux white paper with significant additions regarding software
choices (GNOME over KDE, MySQL, Apache2, Postfix, Python, ...), support
options, and more. "These tasks take money, thus I propose a
membership organization for the service providers (the 'service provider
organization'), that would grant them 'official' status and referrals from
our global service phone number in exchange for their meeting our technical
standards and making a financial contribution. Financial contributions
would be on a sliding scale based on the size of the company, and would be
in two forms: a straight membership fee, and a percentage of new business
referred by the service provider organization." The new text has
been nicely highlighted for those wanting to get a quick idea of what has
changed.
(Log in to post comments)
The desktop issue Posted Dec 10, 2003 11:27 UTC (Wed) by mjr (subscriber, #6979) [Link] I agree with Perens about the choice of GNOME as a default desktop (as well as his given reason for it), but I would've liked, as a matter of principle, the reasons to include GNOME's better focus and support for accessibility. While this isn't a big issue for most users, taking the needs of certain minorities into account should be considered important in itself for an operating system aspiring to wide and general use. Oh well, as long as the right choice was made anyway ;)
The desktop issue Posted Dec 10, 2003 12:54 UTC (Wed) by BrucePerens (subscriber, #2510) [Link] That was discussed on the mailing list extensively and you can read the archive. I didn't want to further fan the flames.Bruce
A new UserLinux draft manifesto Posted Dec 10, 2003 11:42 UTC (Wed) by fjf33 (subscriber, #5768) [Link] Isn't Mozilla Thunderbird a mail client? I think he meant Mozilla Firebird.
A new UserLinux draft manifesto Posted Dec 10, 2003 12:55 UTC (Wed) by BrucePerens (subscriber, #2510) [Link] Right. Sorry.
A new UserLinux draft manifesto Posted Dec 10, 2003 11:43 UTC (Wed) by iker (subscriber, #5011) [Link] Why MySQL instead of Postgres? Seems to me that Postgres has better enterprise features (replication, stored procs, etc).
PostgreSQL Posted Dec 10, 2003 11:53 UTC (Wed) by ncm (subscriber, #165) [Link] That was my first thought on reading the summary. All the other choices are readily defensible.PostgreSQL is so much more mature and capable than MySQL, it seems an obvious choice.
PostgreSQL Posted Dec 10, 2003 15:20 UTC (Wed) by skvidal (subscriber, #3094) [Link] What about the mysql is gpl'd php is bsdish problems of using mysql and php together on a system?That seems like it could cause all sorts of problems that using postgresql instead would solve.
A new UserLinux draft manifesto Posted Dec 10, 2003 12:12 UTC (Wed) by sandy_pond (guest, #9734) [Link] I agree also that PostgreSQL should be included. But MySQL should be also. There's no reason both shouldn't be since there are differences (both license and technical).
A new UserLinux draft manifesto Posted Dec 10, 2003 13:50 UTC (Wed) by leandro (guest, #1460) [Link] PostgreSQL should be included. But MySQL should be also. There's no reason both shouldn't be since there are differences You missed the point of UserLinux. The idea (as opposed to plain Debian) is to make these choices so that the user doesn't need to. PostgreSQL is just superior in nearly every aspect, so there is no reason to include MySQL.
A new UserLinux draft manifesto Posted Dec 10, 2003 15:39 UTC (Wed) by dlang (subscriber, #313) [Link] Postgres is superior, but it is different and people will want to run apps that support MySQL and not postgres so MySQL is needed. however enterprise class products do need features that MySQL doesn't support so they are both needed.
A new UserLinux draft manifesto Posted Jan 6, 2004 6:16 UTC (Tue) by leandro (guest, #1460) [Link] > Postgres is superior, but it is different and people will want to run apps that support MySQL and not postgres so MySQL is needed. For the sake of sanity, these apps should be ported and profit from PostgreSQL's capabilities. MySQL should adapt or die, which it is going to anyway as the official upgrade path to it seems to be SAPdb under the MaxDB name.
A new UserLinux draft manifesto Posted Dec 10, 2003 12:28 UTC (Wed) by stumbles (guest, #8796) [Link] Well I disagree with the choice of Gnome, Debian, MYsql and Mozilla. I don't use anyof them. Would rather see Postgresql and just about anything other than Mozilla. Debian, never cared for. They are going to do the same thing as Redhats treatment of KDE, make it a step child. Which amounts to letting it fend for itself.
A new UserLinux draft manifesto Posted Dec 10, 2003 12:44 UTC (Wed) by TimCunningham (guest, #10316) [Link] What's a (reasonable) alternative to Mozilla?
A new UserLinux draft manifesto Posted Dec 10, 2003 13:50 UTC (Wed) by proski (subscriber, #104) [Link] I think Mozilla Firebird is an alternative. Its goals align nicely with the goals of UserLinux. It integrates with the rest of the GNOME desktop. It's just as stable. It's the future Mozilla. Unless UserLinux wants to encourage writing extensions for Mozilla, Firebird is a better choice in my opinion. Maybe Bruce underestimated stability of Firebird.Mozilla : Firebird :: Debian : UserLinux
A new UserLinux draft manifesto Posted Dec 10, 2003 13:54 UTC (Wed) by leandro (guest, #1460) [Link] >What's a (reasonable) alternative to Mozilla? Galeon or Epiphany.
A new UserLinux draft manifesto Posted Dec 11, 2003 7:42 UTC (Thu) by BrucePerens (subscriber, #2510) [Link] What about plug-ins?
Plugins in Mozilla-based browsers Posted Dec 12, 2003 0:30 UTC (Fri) by mlei (guest, #17766) [Link] AFAIK, Mozilla Firebird, Galeon, and Epiphany all use the same plugin system as Mozilla, namely, drop a .so in the plugins directory. I'm a regular user of Mozilla Firebird and Galeon, and I have had relatively few problems getting plugins to work. It seems that most of the problems people have getting plugins to work is a matter of the plugin being compiled with a different version of gcc or different options, causing incompatibilities. If one would be packaging plugins for distribution, it's a fairly simple matter to compile the plugin so that it works with a given Mozilla Firebird/Galeon/Epiphany package.So, it seems that any of these choices would be reasonable, providing that plugin vendors such as Macromedia and Sun would cooperate in making working packages. Mozilla is probably not a good choice. All three of these alternatives, being Mozilla-based, are just as stable. Mozilla Firebird has a cleaner, very well designed UI. Galeon and Epiphany have the advantage of being GNOME-native. It seems there is really no reason to choose Mozilla, especially because it brings in Mozilla MailNews, and the UserLinux default mail client may well be something else like KMail, Mozilla Thunderbird or Evolution.
A new UserLinux draft manifesto Posted Dec 12, 2003 10:02 UTC (Fri) by mjr (subscriber, #6979) [Link] My experience with Debian (Sid) is that the Mozilla plugins available at /usr/lib/mozilla/plugins are automatically used by Galeon also, and I notice it is the same with Epiphany (I am currently looking at an embedded Excel spreadsheet in Epiphany via the mozilla-bonobo plugin and the Gnumeric component). Thus this shouldn't be a reason to discount these browsers especially given the choice of Gnome as a desktop environment. There may be other reasons to prefer Mozilla, though, such as the fact that Galeon and Epiphany don't provide user interfaces for all of Mozilla's functionality. This may or may not be desired in the default browser of choice.
A new UserLinux draft manifesto Posted Dec 10, 2003 13:04 UTC (Wed) by lutchann (subscriber, #8872) [Link] I'd say KDE has fended for itself quite well so far.I've never been a fan of KDE because the GUI felt like it was just hacked together without any forethought. GNOME 2 feels like a much more well-designed GUI, and as far as I can tell, doesn't lack anything that KDE has.
A new UserLinux draft manifesto Posted Dec 10, 2003 13:16 UTC (Wed) by stumbles (guest, #8796) [Link] Lol, those are the very same comments I have heard others say about Gnome. Sothey ya have it and that's the rub.
A new UserLinux draft manifesto Posted Dec 10, 2003 13:57 UTC (Wed) by proski (subscriber, #104) [Link] Did you read the manifesto at all? The idea of UserLinux is to make it easy to write non-free software for it (and distribute is too, I assume), and KDE doesn't allow it because it uses libraries licensed under GPL.
A new UserLinux draft manifesto Posted Dec 10, 2003 14:24 UTC (Wed) by hazelsct (subscriber, #3659) [Link] I am a long-time GNOME user and fan, but this logic seems rather self-contradictory. First, Bruce laments the sorry state of the commercial distros: that they charge per-seat license fees, and often include proprietary software. Then he says that we need to use GNOME for UserLinux so people can build proprietary software for it (and also says that we should encourage service companies which charge per-seat license fees, but that's off-topic for this thread...).What gives? Isn't this why RMS renamed the LGPL "Lesser", and encourages the use of the GPL instead? Bruce, are you for Free Software, or for proprietary software? Developers can also build (and sell) proprietary software for KDE too, they just need to pay TrollTech a bit of money, which goes back into the development of the (free) Qt toolkit.
A new UserLinux draft manifesto Posted Dec 10, 2003 14:40 UTC (Wed) by johnny (subscriber, #10110) [Link] I think it would be nice if the target was to make it easier to writefree software instead of non-free software. KDE/Qt certainly makes it easier to write free software than Gnome/Gtk. Oh well...
A new UserLinux draft manifesto Posted Dec 10, 2003 13:08 UTC (Wed) by ninjaz (guest, #2083) [Link] The idea behind this proposal sounds good, but I am leery because it calls for using a Debian base.The fundamental problem with Debian is there is no concept of "core" like classic Unix operating systems, or, for instance, *BSD. This means updates to support new hardware, integrate new (and often much-needed) features in the kernel and libc are pushed back until a release far in the future, because they're waiting for 2000 user application packages to stabilize first, too. I think having a regularly released Debian core (every 6-12 months, maybe?) with user packages maintained separately would be much more desirable than the current situation of what has been referred to as Debian -broken and Debian -obsolete. IMHO, user application packages would be better decoupled in a similar manner as FreeBSD ports so they don't hold up releases. I don't want to wait for gtkMySQLoggDJ have its critical bugs closed before I can get the core system updated on my servers.
A new UserLinux draft manifesto Posted Dec 10, 2003 15:12 UTC (Wed) by hazelsct (subscriber, #3659) [Link] I have to disagree with a couple of points you've made.First, the reason Debian releases every two years, not every two months, is not because of the time required to stabilize thousands of packages, but because it takes that long to (attempt to) make a new working installer. Why does it take that long? Because Debian is a developer-run volunteer organization, and developers already have Debian installed, thus not many developers are interested in hacking the installer. Second, IMO Debian's release cycle is a *good* thing, in a couple of ways. I run a research group of grad students, and though they're not free software hackers, they do code, and even they appreciate having a consistent environment for the last year and a half since the woody release. Corporations do not want to re-train their entire workforce every three or six or even twelve months. Which is why Microsoft releases only every two years or so. At least as importantly, some of the software does only stabilize about that often. Woody had GNOME 1.4, which was a mature, stable product. Since then, RedHat 8.0, 8.1, 8.2, 9, and Fedora Core have been released with varying degrees of GNOME 2 maturity, and ugly mixtures of GNOME 1 and 2. Debian Sarge will release with a very solid, mature and complete GNOME 2.4 or 2.6 system. From 1.4 to 2.4, from strength to strength, with consistently stable applications. So when the new installer is ready, Debian will then be able to release more often. But I'm not sure that's a good thing, my group and I rather like the current release pace, and IMO enterprises do too. If you don't, well, that's what testing and unstable are for, or use another distro; but I don't think you can make the case that enterprise users, admins and trainers really want to upgrade their distribution every few months.
A new UserLinux draft manifesto Posted Dec 11, 2003 9:13 UTC (Thu) by tjc (guest, #137) [Link] > Since then, RedHat 8.0, 8.1, 8.2, 9, and Fedora Core have been released3 out of 5.
Freedom of choice? Posted Dec 10, 2003 13:26 UTC (Wed) by southey (subscriber, #9466) [Link] One of the superior aspects of Linux over all other operating system is the freedom of choice. This white paper is a major step backwards in creating a user orientated Linux distribution. Further it doesn't consider vendor's needs - too bad if that package needs perl or whatever. If a package that you need is not carried in your distro then you will switch distros very quickly to avoid needing to reinstall the package and keep the associated user settings.It's far better to have all the major packages and basic libraries in place from the start - then you can run Gnome apps under KDE or use that perl/C/java code. 'Look and feel' is irrelevant if the user knows what they are doing.
Freedom of choice? Posted Dec 10, 2003 14:07 UTC (Wed) by proski (subscriber, #104) [Link] An interesting aspect of the problem is how UserLinux is going to handle future releases. Today's underdogs will be tomorrow's mainstream. What if (random example) Ruby makes Python obsolete? Will UserLinux stick with the choices made in year 2003 or will it switch to the next big thing?Choice between mature and emerging technologies is essential for healthy development. That's the kind of choice that shouldn't be eliminated, or the system will stagnate.
restricting choice is BAD. Posted Dec 10, 2003 14:51 UTC (Wed) by pizza (subscriber, #46) [Link] There's one big thing that everyone seems to be missing -- it't not about too much choice. Arguably windows has just as much choice, if not more, when it comes to available software. it's just not bundled with the base OS, and you don't get source code.Instead, what's needed is the ability to highly tailor isntallations with relative ease, which is why RedHat is liked so much -- Anaconda's kickstart facility lets you stick a floppy in a machine, come back an hour later and you have a completed install tailored to a specific configuration, complete with a customized software group/package list. This is especially important when you consider that they're targetting both the "server" and the "workstation" market with a single "UserLinux". The two markets have very different requirements, as far as what a "base system" comprises. The "UserLinux Core" idea is excellent; when it comes down to it only a standard set of libraries and syatem facilities need to be installed for ISV support and certification. But arbitrarily drawing a KDE/GNOME line is silly; what if some ISV built their app with QT instead of GTK+? Just because you're not going to use KDE as the "default" desktop doesn't mean that the libraries shouldn't be installed (optionally) so Oracle15r2 will work without any external libraries? So Big, Inc's IT department will build a customized install, with it tailored to their requirements, not the arbitrary decisions of the UserLinuxConsortium. Remind me, again, how that would be better than having the "take it all or leave it" choice of Windows forced upon you? Remember, the "User" in "UserLinux" refers to the IT department, not the secretary who is using the computer. For "small business" (ie one without a dedicated IT department) the overhead of doing custom installations is less desirable; however one could just consider that another install target "Default Install For Small Business Desktop". But I wonder if going with a recent off-the-shelf distro might just be a better choice for those people, as they're going to install everything just in case, or compare labels on the software they need and make sure the distro they install has the necessary features. The application [software] drives the OS choice, not the other way around. Businesses ask the question "what do I need to do X", and they go with that. Linux's biggest strength is its flexibility and choice (well, and Free Software too..) I really wish people would stop prattling on about the upfront dollar cost, because nothing is free in that sense. Quite frankly, it looks like UserLinux is trying to copy the balance that RH/Fedora already has -- a balance that does not fit in well with Debian's strengths.
restricting choice is BAD. Posted Dec 10, 2003 15:32 UTC (Wed) by pizza (subscriber, #46) [Link] Great, I follow up to myself. Heh.I must say I'm confused; is UserLinux's goal to be friendly to IT departments (ie deployment/operational support), IT/ISV developers (eg the likes of Oracle or other commercial software vendors), or friendly to users? Or all of the above? For the first two, choice is a damn good thing. For the latter, choice is not so desireable; insofar as the target application/tasks determine what other stuff the end-user will need on their machine. The end-user's wants are largely irrelevant. Hate to bring Windows back in; but Windows is a fixed target with no "choice" -- but buy MSDN for the developers, and you essentially get "everything else" thrown in for good measure, and I do mean everything... including the files you'd need to redistribute with your app to get them to work on the "bare" target machines. (and MSDN includes several scripting languages, several different Databases, etc, etc, etc...) plus, for the IT operational support, MS has also provided full scripted customized installs since at least the NT 4.0 days. I'm just confused as to what problem UserLinux is really trying to solve. Things like the installation crap that I harped on earlier are easy to solve, within the main Debian framework, even.. so it's really moot. I'm just concerned that arbitrarily restricting _developer_ choice would be detrimental to getting Linux adopted in the first place, as it's like shooting one of Linux's greatest strengths in the foot. Third-party proprietary software companies need a stable (yet reasonably up-to-date) platform to build on. RH's managed to pull this off quite successfully, but they're charging money for it. But pulling it off costs a non-trivial amount of ongoing money, first to build it, and then to actively maintain and support it for however many years as necessary. For example, Win'98 was supported for close to five years!
No choice is what keeps Windows in 90% of desktops. Posted Dec 11, 2003 16:40 UTC (Thu) by khim (subscriber, #9252) [Link] You can think about it as really, really, really bad thing but most end-users like lack of choices. Not as much lack of choices by itself as they like ability to use thing without knowing such choices even exist!
Choices are scary. That's exactly why I still do not use Debian: installer keeps asking me about things I know nothing about ("do you want to use xyz for some_obscire_task or abc?" "@Q$^&#$@! how the hell should I know when I never seen neither xyz nor abc?"). And I'm not Windows newbie. Now if 100% choice-less UserLinux will be the only distribtion around - it'll be disaster. But as "distribution for masses" it's good. They do not need nor want choices. At least not deep at system level. "Do you like you desktop blue or green?" is probably already too complicated for them, let alone "do you want to use KDE or GNOME?".
A new UserLinux draft manifesto Posted Dec 10, 2003 15:00 UTC (Wed) by pointwood (subscriber, #2814) [Link] I like it, I think it is a really good idea!I generally prefer KDE, but I can understand the reason for choosing Gnome. I think both should be included though. With all the interoperability work done in freedesktop.org, I don't think that is going to be that big an issue in the long run either. You should also consider that KDE pretty popular, especially in Europe (AFAIK), I think it would be wise not to "exclude" them. I'm sorry that I bring the issue up again...
A new UserLinux draft manifesto Posted Dec 10, 2003 15:44 UTC (Wed) by proski (subscriber, #104) [Link] From the manifesto:I think it makes sense for an enterprise project to make choices among the two complete GUIs available on Debian, the dozen web servers, and so on. Having a bounded set of packages to collectively support and improve is important, especially at the beginning.I don't think interoperability is relevant here. Mail servers are compatible and all of them can emulate sendmail, yet UserLinux will support only one mail server.
A new UserLinux draft manifesto Posted Dec 11, 2003 2:20 UTC (Thu) by pointwood (subscriber, #2814) [Link] I know, I did read the manifesto before posting ;)As I said, I understand the reasons for choosing Gnome and only one desktop solution, but I don't agree and still think it would be unwise to exclude one of them.
scripting languages Posted Dec 10, 2003 23:36 UTC (Wed) by dlang (subscriber, #313) [Link] Bruce, why do you feel you need to declare a single one the winner?you can try to say that you have one that's prefered for new stuff, but trying to say that all programs written in perl will need to be re-written in python (or ruby or X) seriously complicates things. are you also going to declare a single shell as the only one on the system? yes you definantly don't want to support a dozen MTA packages, but for areas that you have several large groups useing a relativly small number of choices, choose all of them. your total package count will still be a tiny fraction of the available Debian packages and you will avoid internal wars over the tools. for the KDE vs Gnome debate, personally I don't use either, but I still have them both installed on my desktop machine so that I can use programs written for either one. Unless there are fundamental incompatabilities between them install both, go ahead and pick one as the default, but don't try to ban the other from the distro.
scripting languages - and other choices Posted Dec 11, 2003 3:57 UTC (Thu) by dwalters (subscriber, #4207) [Link] I totally agree with this point. You've just got to support Perl (and PHP for that matter) as well as Python - there's just too much stuff out there based on it (who can live without SpamAssassin these days?)There doesn't have to be just one choice for everything. Let's change the manifesto a little to accommodate this fact. For the same reason, I believe UserLinux should support both PostrgreSQL (a far superior enterprise-class database server than MySQL), and MySQL as well. MySQL is "the Perl of SQL databases", if you will. In other words, there's a whole slew of LAMP applications that have been written that depend on MySQL, so UserLinux needs it for that reason, but it's still the wrong choice if you are going to select just one RDBMS for an Enterprise distro (which UserLinux is supposed to be). Of course it makes sense to make choices as far as the user's desktop experience is concerned, but as far as the back-end stuff is concerned, I think that choosing between different shells, scripting languages and servers is unnecessary and makes UserLinux less, not more appealing to Enterprise customers. You may notice that I included shells in that list, even though Bruce's white paper doesn't actually mention anything about choosing bash over any other shell. Perhaps this is because Bruce has no intention of choosing any one shell over another. Well Bruce, if that's is the case, you must ask yourself why you feel the need to choose one scripting language over another (and RDBMS too).
scripting languages - and other choices Posted Dec 11, 2003 7:52 UTC (Thu) by BrucePerens (subscriber, #2510) [Link] Obviously we can't remove perl from the system, Debian packages (some of which include perl installation scripts) and any number of programs would break. However, we can say what language we encourage people to write in - both for the project, and for our customers.Designating one product to go with is about managing the support burden. One of the problems we had at Pixar was that technical directors had to know 19 different computer languages - some special-purpose ones that we'd developed, all of the various scripting languages we'd used, etc. Reducing the burden on support staff is very important. Getting a focus on one GUI, scripting language, database (OK, everyone says PostgreSQL is the right choice), etc., means that support staff have a managable burden. If customers want to depart from that list, they have the right to do so and the packages are in Debian. They might end up paying a premium for service, but I think we can let the market deal with that problem. Bruce
scripting languages - and other choices Posted Dec 17, 2003 5:49 UTC (Wed) by mwilck (guest, #1966) [Link] You may notice that I included shells in that list, even though Bruce's white paper doesn't actually mention anything about choosing bash over any other shell.And he didn't say anything about choosing vi over emacs (or vice versa), either ... :-)
Limiting choices Posted Dec 12, 2003 0:48 UTC (Fri) by mlei (guest, #17766) [Link] It seems that all Bruce is proposing in choosing one product from each category is to make that the one that will have to be supported by companies providing support for UserLinux. "Reasonable defaults" are the idea here. Being based-on-Debian will mean that a customer can install those other programs that they need if they need them. If they really need those programs, they can support them themselves, or pay someone to do that extra support.This isn't about removing choice for users who want it. It is about not presenting choice to users that don't want it. I also think (and correct me if I'm wrong, Bruce) that he isn't necessarily calling for excluding KDE libraries from the system. If an application needs them, they should be present. But the default desktop will be GNOME, so a user won't be presented with the choice of booting into a completely different DE. Supporting multiple UIs is somewhat of a bane for support staff. I worked on a university IT department last year and the first question always had to be "which version of Windows are you running?" And so there were three or four different answers for each question, depending on if it was 95/98/ME, NT, 2000, or XP. It's much easier to work with one UI. It saves time for the customer and the support staff, which translates to more work done at less cost.
|
|||||||||||