Welcome!

Linux Authors: Dana Gardner, Elizabeth White, Gary Kaiser, Noel Wurst, Liz McMillan

Related Topics: Linux

Linux: Article

Linux.SYS-CON.com Exclusive: What Would UserLinux Look Like?

Bruce Perens tells LinuxWorld's desktop editor what he has in mind with UserLinux

Last Monday at the Desktop Linux Consortium Conference at Boston University’s Tyngsboro, Massachusetts Campus there was a lot of talk about a “UserLinux” distribution. The topic was sparked by remarks by Bruce Perens who voiced a need for a distribution that was designed to meet community needs for a desktop operating system based on the Linux community favorite Debian distribution.

I contacted Bruce who has been kind enough to interject some comments to my own text. They are marked [thus].

The thought of UserLinux sparked my thinking. The thing I like about Linux is that it’s infinitely customizable to meet the needs of almost any situation. However, for it to be a viable desktop for the masses there seems to me that there has to be some common features that a large number of Linux desktop users would appreciate. I thought about this quite a bit and started my list of what it would take for Linux to be my “ideal” environment rather than my preferred environment. I’d be interested to see what the community considers the most important features.

[Bruce Perens writes: I should point out that UserLinux also has a server mission. Our first customer group has both server and desktop needs. But the server is a solved problem, at least mostly, so we know a lot of work needs to go into the desktop.

Also, the most important thing about UserLinux is that it is an attempt to change the economic paradigm of the Linux distribution. We feel that creating a Linux distribution doesn't work as a profit-center, and that it is better viewed as a cost-sharing exercise. So, the customers involved in UserLinux will be paying for the engineering of creating a Free Software system, rather than for boxes, "seats", or user licenses. The system will be certified to various standards and vendor requirements with their funding, and the result will be given away. The customers get all of the copies they need with no incremental cost per seat added. They will have to pay for service.]

My list has two overwhelming requirements for the Linux desktop. First it has to be easy to use. It should pass the “Grandma test” which is when placed in front of the average grandma she would find it intuitive and easy to use. Second it should include a set of tools that allow the user to accomplish their most important tasks. I generated my list of tools and what I feel are my most important for my needs. I would encourage you the prospective users of such a system to add your feedback.

 

Productivity Tools

Browser ­ I think Mozilla is a great option for browsers. I like the tab-based browsing and pop up blocker. If not Mozilla than maybe some of the projects spawned from Mozilla aimed at speedier performance without the frills like Firebird.

[Bruce Perens writes: I'd like to hear if Konqueror has something to offer that is not matched by these choices.]

Office Suite ­ I use Open Office and Star Office and I think they are good. For some of my more ambitious projects I do use Microsoft Word but I find myself using Microsoft less. I particularly like the ability to export files to PDF format preserving the look and feel of my files across platforms. If these suites could handle better more complex formatting I think they would easily displace their competitors that costs many hundreds of dollars.

[Bruce Perens writes: I like OpenOffice and hope that I can facilitate the creation of a broader development community outside of Sun.]

E-mail/PIM ­ Outlook made the integrated PIM and email client the vogue in business. I like the idea but I think that Microsoft’s implementation is lacking. So far the best Linux solution for me is Ximian Evolution but it lacks some features I like about Outlook. Particularly the ability to drag e-mail messages to a task list or calendar. In Ximian’s favor is the RSS integration into their Summary page to gather my news all in one place. Once again this is a case that I primarily use Outlook running on a virtual Windows environment Win4lin.

[Bruce Perens writes: Well, when there are features lacking in an Open Source program like Evolution, you know what to do, don't you? I think that a solution to the ones you complain about could come from the community.]

Financial Software ­ I use Quicken and TurboTax mainly because I have for years and I think they are both very good products. I know GNU Cash (www.gnucash.org) is an option and I am actually playing around with it right now but it will be a hard move for me. Not only because of differences in features but the learning curve.

[Bruce Perens writes: I haven't looked at these closely yet. I actually still have one Windows machine in my home, and need it for TurboTax. I still have Quicken on it, but think I could move off of Quicken if I had to.]

 

Utilities

Application Installation ­ This is probably my biggest complaint with most Linux distributions. RPM installation often results in dependency problems. Causing me to search for the recommended libraries to fulfill dependencies so that I can install my application. Debian’s apt tools and apt4rpm both work very well making things easier for most users. However, many of the most popular distributions still use plain old RPM warts and all. I think that a good one click install like available through Lindows Click N Run Warehouse would be ideal for ‘User Linux”.

[Bruce Perens writes: The solution here is obviously some front-end on top of apt, and Debian packages. It's really strange that people still

complain about RPM dependencies, I don't understand why Debian was able to solve this so many years ago and Red
Hat still has a problem.]

Docking and Power Management Tools ­ For laptop users like myself I find that most distributions don’t handle hot docking and undocking of laptops well. In my Utopian Linux distribution I would want to see the ability to “hot” dock and undock my laptop by clicking a button.

[Bruce Perens writes: You shouldn't have to push that button. You should just be able to dock and undock. But Linux ACPI is still immature,
and is not going to be in a good state for most laptops with the release of kernel 2.6 . I spoke with Dirk Holmdel of Intel about this, he feels that the present Linux ACPI drivers don't handle all of the start-up and shut-down tasks in the right order. Also, most kernel drivers have not been ported to the new driver model yet, and do not handle power management correctly.
I have a problem with various laptop graphics chips and wireless chips, because their manufacturers are unwilling to document them fully. We might have to start publicizing a "not ready for purchase" list for various hardware manufacturers that can't get with the program. I think that even Windows customers will be reluctant to purchase a laptop that could not ever be switched to Linux.]

Backup Utilities I have the expertise to set up cronjobs that rsynch my desktop to my file server but most people don’t. I would think client-side tools to synch files to file servers of all types would be a welcome inclusion. This tool would be make it easy to schedule backups and choose files for backing up from an intuitive interface.

[Bruce Perens writes: It would be interesting to see if some of the disconnected filesystems like Coda could help with this. Potentially they remove the need to consciously synchronize things. Just dock and it gets done.]

Windows Networking Client ­ The majority of businesses I go to today use Microsoft Windows Server for file and print sharing. Having the ability to browse these networks would make things more convenient for me. I often use LinNeighborhood, which is an easy to use Windows network browser. I think overall platform interoperability is the key to Linux adoption.

[Bruce Perens writes: Yes. Since this is a solved problem in the free software world, it should go into the system.]

I could go on for days about my ideal desktop but what I am curious to know is what’s your ideal incarnation of Linux desktop. Maybe we can point your feedback to Bruce as he works on his proposal to help shape his proposal for UserLinux.

[Bruce Perens writes: I am also interested in knowing what people feel is missing from the server.]

 

More Stories By Mark R. Hinkle

Mark Hinkle is the Vice President of Community at Cloud.com. the maker of the open source cloud computing management software, CloudStack He also is along-time open source expert and advocate. He is a co-founder of both the Open Source Management Consortium and the Desktop Linux Consortium. He has served as Editor-in-Chief for both LinuxWorld Magazine and Enterprise Open Source Magazine. Hinkle is also the author of the book, "Windows to Linux Business Desktop Migration" (Thomson, 2006). His blog on open source, technology, and new media can be found at http://www.socializedsoftware.com.

More Stories By Bruce Perens

Bruce Perens, a leader in the free software and open source community, is a member of the International Advisory Board of Linux.SYS-CON.com. He is the creator of the Open Source Definition, the manifesto of the open source movement. Bruce is founder or cofounder of the Open Source Initiative, the Linux Standard Base, Software in the Public Interest, and No-Code International. He is the creator of Busybox, which has spawned its own development community and is part of most commercial devices using embedded Linux.

Comments (134) View Comments

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.


Most Recent Comments
Michael H 11/19/03 12:40:09 PM EST

Financial Software:

I've heard the Quicken vs. GNUcash argument over and over again. I've spoken to the GNUcash folks at shows, and every time I tell them what I see is lacking in GNUcash vs. Quicken. They usually agree that it would be nice, but it never seems to make any progress.

Whether you use GNUcash or Quicken, the *BASIC* functionality is no different than a simple spreasheet: track my expenses, track my income, report. I could do that now with vi, awk, sed, and join.

What differentiates Quicken from CNUcash, spreadsheets, or even vi is their add-in services, mostly around the network stuff. I pay all of my bills electronically via Quicken. I don't need extra software; I just enter who I want to pay in the ledger and set the "type" to "send". I hit a button to update (connect to my Bank's bill paying service) and my electronic payments are sent, a serial number is assigned to them which makes it's way back into my register, and I can pretty much forget about it. If I have trouble later, I can "inquiry" a payment and even send a message to the bill paying service and they will track down why X company did not credit an electronic payment they sent them on my behalf.

But that's just part of the value.

Next, I can download my latest cleared transactions from my bank, as well as what they believe is my latest balance. Quicken uses this information to automatically reconcile my checking account every time I download (daily, weekly, whatever). Problems are identified LONG before my statement; in fact, I don't even use my statement anymore... I'm reconciled every week. The statement just serves more or less as a record of payment. So, I spend zero time each month reconciling my checking account.

My "one" Windows machine at home will remain, running Quicken (and oh yeah turbotax) until someone figures out that "accounting software" != "glorified spreadsheet".

Harry LeBlanc 11/19/03 12:28:34 PM EST

You've overlooked one crucial component. Many small businesses have invested in custom applications running in MS Access. Yeah, yeah, I know that Access isn't a *real* database engine, but the fact remains that many small businesses -- who are otherwise prime candidates to jump to linux -- are tethered by their custom Access applications.
So high on my wish list would be a good desktop database front-end, with forms, reports, ad-hoc queries, and easy scripting --- *AND* a way to migrate Access applications.
I know Rekall just created a GPL version, and there's GAMBAS, a BASIC not unlike VB. If someone would write a conversion routine that converted most of the components, there would be a gold rush of small business conversions.

Ken R. 11/19/03 12:01:43 PM EST

I think the only things that need to be added are a simple installer that auto-detects your hardware and installs the correct drivers (lots of distros are working on this including debian/Gnu); interoperability with other operating systems; a simple update like Ximian built for Red Hat.

For Linux to gain acceptance, it's going to have to see the addition of some killer apps. I know that it has a lot already but Linux needs to be an innovator not an imitator when it comes to desktop apps.

Bruce should put a Wiki application up on http://www.userlinux.com so we can organize all of these great ideas that people have contributed so far.

Aaron B. 11/19/03 11:46:05 AM EST

pls. note: the above comment originally had angle brackets around the first line reading:

{{{{{no disrespect intended but...}}}}}

apparently the parser thought I may have been guilty of some XSS...

Aaron B. 11/19/03 11:44:29 AM EST

It's the mouse, stupid!!

Yes, I use linux on a daily basis. (as well as windows) And the single most annoying thing is that the mouse dosen't move like I expect it to. Acceleration, speed, etc. I've tried to tweak it with little success, and it drives me nuts.

Ok, ok, I readily admit that this may be a user issue - but I think it's reasonable to expect that I can calibrate the mouse "feel" to be reasonably close to where I'm expecting.

Or maybe I'm just too picky. =^)

Gordon Gray 11/19/03 11:31:08 AM EST

I agree with Dominic. Having managed dozens of projects over the past 20 years, I find that most (not all) programmers would enjoy being given directions on how it should work as much as inventing how it should work. If an HUI person is doing the giving, then the programmers usually can buy into it.

Dominic Chambers 11/19/03 11:14:58 AM EST

To be honest, I think that if your were to select the best parts from a number of existing distros you would already have a stellar distribution. With a little extra polish it could be the seminal modern desktop. Collectively, ArkLinux, Suse, Xandros, Lindows, et al have already nailed this problem; it just needs integrating.

A lot of the comments above are great, but be careful with trying to please everyone or we will end up wih buggy, inconsistent slops. For example, choose KDE OR Gnome as the base, or you'll never end up with a tightly integrated system.

Finally I would like to make the most important point of all. No active developers should be able to set policy on how this will look. We get to close to the problem and screw it up. Instead, HUI and designer types should mandate how it will work and the developers should just build it.

Best of Luck.

Luke 11/19/03 11:01:23 AM EST

I think that Linux as a desktop has greatly matured in the last year alone, however there are allot of things that I would like to see changed or improved. First, let me say that allot of features that people complain about are all ready in place. It just depends on the distro that your using. I found that redhat is by far the biggest pain in the @** distro out there. I've switched to SuSE on my desktops and Debian on some servers.

I would like to see tighter kernel security management made easy. I can't tell you how many times I've stayed up until 4 am kernel tuning(not fun!).

Also, the ability to play different types of media(namely DVD). I've found that SuSE is pretty good for this although a little bit of tweaking is necessary.

I think that Platform integration with windows is key to becoming a real competitor for MS. If I throw a windows app cd in my Linux laptop, something like WINE or Win4Lin should pick it up as soon as I click "Setup.exe". I know that this is kind of frivolous but would be nice all the same. I also see allot of major Software vendors stepping up and porting their stuff to Linux.

Conversely on the server side of things, I need management tools. Enterprise management tools. I've been really impressed with Novell's ZENWorks and Ximian's Red Carpet Enterprise. However I'd like to see something like this in every server distro.

On the negative side there are some things, I think, that need to go away, or go in the background. Kerberos. Stop that, just STOP THAT. It's OLD technology. If I wanted to run my enterprise on System390 and AS/400 I'd run kerberos. LDAP is the wave of the future, accept it. This statement is largely pointed at Redhat, which until recently would only use kerberos authentication with any ease. LDAP was a nightmare to configure, and when it was configured it didn't work right. Now with Fedora they have SMB authentication(::yeah::). I understand that kerberos is in use by a lot of ISP's, but that doesn't mean it's good for regular networks. Also, the whole clustering thing. What for? I understand how fantastic is it to take 200 4x86's and cluster them into your brand new supercomputer, that serves no real function. It's just one of those things I think should go into a special distro or package.

That would seem to be it, aside from all of the things all ready pointed out. I think that Linux has truly grown, and will continue to.

Jasper Nuyens 11/19/03 10:29:37 AM EST

Ali Naddaf wrote: "I believe this feature, the way it was stated, is indeed a killer feature :-)"

I was talking a little ahead of current implementation now... Reiser is searching for a sponsor to have it implemented in reiserfs :-)

The slashdot interview:
http://interviews.slashdot.org/article.pl?sid=03/06/18/1516239&mode=flat...

I think it could be a WinFS killer :)

John M 11/19/03 10:29:17 AM EST

I would start with the simple functions I can't accomplish today. I would like to be able to plug my laptop into the net and have it automagically DHCP and address (link monitoring).

I would like to be able to plug in my USB dongle, edit a word document on it, remove that USB dongle and not have to worry about modules, filesystems, mounting, mountpoints or anything else... GNOME/KDE _almost_ have CDs and floppy disks just in time for 'em to become obsolete. Now we need the 50-port all-in-one-read-any-kind-of-memory-stick flash readers.

Yair Carel 11/19/03 10:23:37 AM EST

First of all the focus point should be on non american users because Europe and many asian countries are ready for a change from the monopolist world of closed source programs (especially microsoft), and also because it has a lot to earn from decentralizing the software industry from the Mega corporates (mostly american software companies).
When Open source (and linux particulary) will be widely deployed all over the world the american market will follow.
What this means is that EVERY linux distribution should have:
- a TRIVIAL internationalization features:
1. Make the system 100% Unicode based, with only one encoding for all programs and systems.
2. Have a perfect BIDI support so the same system can be used in every country and by everybody with no special settings done. (KDE is almost there so KDE is a must).
3. The most urgent need is to transfer Evolution to the QT API, so it has the best BIDI support.

- Laptop support has to be perfect because many users are dumping a desktop for a laptop especially in the corporates where workers are more and more mobile.

- Leading firewall suppliers like checkpoint have to be convinced to make vpn clients for linux so workers can connect to the company's network.

- A dialer for the AT&T worldwide internet access service.

- Last thing is a good and practical harware managment GUI. probably like the Mandrake hardware wizard but with the possibility to custom install device drivers from binary and source. (this of course implies auto install and managment of kernels).

Matthew Lowrance 11/19/03 10:13:48 AM EST

I agree with most of the comments above, keeping it simple is the most important here. I like the idea of picking ONE app for each task that works and then making installing others easy.

I am more interested in mentioning what should NOT be included. I have tried many of the desktop distros and have always been annoyed to see that once the install finished and I rebooted about 10,000 services started that a "desktop" user will NEVER use. Why is Apache installed by default? Why is mySQL installed by default? The list goes on and on. I know power user want these, I personally want them, but I would be perfectly happy having to install them afterwards knowing that there's not a million computer newbies out there with Apache open on the internet and they don't even know it.

I'm not trying to troll, I'm just trying to suggest this up front, so this (and ALL the other desktop distros) don't run into the same security problems by default. Most of the Windows problems are caused because people don't install patches and just because they move to Linux doesn't mean they'll start (or even that they know they should).

Two last things, first a "Windows Update" type applet that makes it very easy to install important updates and second a basic firewall turned on by default.

Hope this helps, I'll be very interested in following the progress of the project and maybe helping out once it gets started.

Justin 11/19/03 10:00:26 AM EST

Konqueror supports SMB so you don't need linneighborhood just browse in konqueror like smb://192.168.0.101/shared_files or whatever.

Brian Tarbox 11/19/03 09:38:15 AM EST

How about a tool to configure the 'Start' menu? In RedHat 8 or 9 I don't see a tool to do this.

Ali Naddaf 11/19/03 09:15:32 AM EST

Jasper Nuyens wrote "The versioning system of Reiserfs should get some kind of graphical front-end. Even longhorn will not have versioning on EACH AND EVERY file (text file, config file, application datafile). The posibilities to roll back to an older version of ANY file is a KILLER FEATURE available right now with reiserfs. Bring this to the GUI-public fast and we have another MS-bashing reason :-)"

Jasper, I an very interested to learn more about this feature, would you be kind enough to post a reference to this feature? I believe this feature, the way it was stated, is indeed a killer feature :-)

Thanks.

blalkhsadf 11/19/03 09:04:53 AM EST

gdm already has "fast user switching". From the gnome foot menu, select "New Login" or run gdmflexiserver. It will start a second X session. When you log out of that one, you are back in your first session.

Keith Weiland 11/19/03 09:00:15 AM EST

Only 3 things stop me using Linux exclusively, if these were included in any distro I would use it.

1. Confusion over the hundreds of applications installed by default with crazy names and no indication of what they do, finding 20 text editors installed when one best of breed would do. Make a decision and include only one example of any "type" of application on the CD or ISO based on its commercial features and quality. Any other applications should be obtianable through download for those who already have a preference. This also means limiting default desktop to one choice. Anyone who already has a preference will most likely be able to install it later.

2. Get support for the latest, hardware especially on the graphics card front. Negotiate if possible to get the drivers opened up so the community can work with them. If this fails with ATI and nVidia, get XGI onboard and at least we will have easy hardware acceleration.

3. Get some decent Game Designers interested. I cannot play Unreal and Quake sequels forever. Some MMORPGs etc would be great.

If 1 and 2 were taken care of 3 would follow anyway in time. Otherwise what has already been mentioned is a good start. I would just say that in the end a TRUE desktop system is one that will be happily used in the home by the whole family. Getting linux on the Corporate desktop will be easy in comparison.

J. J. Ramsey 11/19/03 08:53:32 AM EST

About Konqueror:

The main advantage that I see it having over Mozilla is that its look and feel is the same as that of KDE, while Mozilla sticks out. However, Konqueror has two weaknesses. However, Konqueror doesn't handle the "corner cases" of the Web as well as Mozilla. For example, uploading to briefcase.yahoo.com works on Mozilla, but not Konqueror.

About RPM and dependencies:

In the Debian world, there is a three-branched primary repository of packages (the branches being stable, testing, and unstable), and any other secondary repositories are expected to be compatible with at least one branch of this primary repository.

In the world of RPM-based distributions, there are several different repositories, most of which are maintained by particular individuals, and there is no guarantee or even necessarily a reasonable expectation that these repositories will be consistent with one another.

Here Debian has an advantage.

About drivers:

In the Windows world, the Windows CD comes with some drivers, but it is expected that third-party drivers (from the hardware manufacturers) will be installed in order to handle most of the up-to-date hardware, and installation of these drivers is fairly trivial--barring bugs, of course.

In the Linux world, the vast majority of hardware drivers come on the CD, while third-party drivers handle the corner cases. Generally, installing third-party drivers involves using the command line and editing config files. This needs to change. For the most part this problem can be solved by packaging the necessary driver files and then running the appropriate graphical config tool to use the drivers.

Kernel modules are a special case, because there is no ABI for them, nor is it likely that Linus Torvalds will change this. Here UserLinux could, and probably should, set a new standard. When a kernel driver package is installed, the internal install procedure should install the source (and any binary parts of the module, if any) to some directory, i.e. /usr/src/kernel-drivers/, compile the module, and install it. When the Linux kernel package itself is upgraded, the post-install procedure should recompile any third-party modules from the source installed under /usr/src/kernel-drivers/, and install the recompiled modules for the new kernel. In short, the package manager would handle compiling and recompiling modules without direct user intervention.

Jasper Nuyens 11/19/03 08:46:39 AM EST

First, greetz from a former VA life and congratz with the initiative :-)

Take a look at KNOPPIX. It accomplishes already quiet a lot of things you aim for; it's debian based and it's hardware detection is FANTASTIC, also on laptops and with different wireless hardware.

What I feel to be very important and neglected a lot is wine.
Wine is getting very good these days and gives the opportunity to 'just include' all windows apps.
MacOS X has different api layers, and wine should be as good integrated in this new system, with working mime-additions and such.
If users feel like installing Word on their Linux boxes, why not (it even runs faster with wine than Open Office). But it should be as easy to install as on Windows (e.g. the autorun should start). Don't hesitate to contact the CEO of codeweavers for collaboration, he's a nice guy.

Importantly, GUI configuration tools should work properly, I hope you don't go the RedHat road here. Any new RH sysadmin gets the advise to always do everything on the commandline.

Fonts have always been important, but neglected in OSes the last 10 years, there exist plenty of free and high quality fonts on the web, yet no distro/OS takes the time to sort out the good ones and include them in their system. Caligraphic/iconic/funny fonts should be standard!

Don't bet on weird distributed filesystems, Linus said it right in one of the recent interviews, it isn't there yet. Go for rsync in the mean time... it's fast, efficient and secure. Link status can be autodected, even with a shell script running ifconfig...

The versioning system of Reiserfs should get some kind of graphical front-end. Even longhorn will not have versioning on EACH AND EVERY file (text file, config file, application datafile). The posibilities to roll back to an older version of ANY file is a KILLER FEATURE available right now with reiserfs. Bring this to the GUI-public fast and we have another MS-bashing reason :-)

DVD/MP3/MEDIA-players should be included by default, just as KNOPPIX does,... if fear from the DMCA, distribute physically from outside the US.

Important for systems with more than one user is also a modified xdm/gdm/kdm to allow 'fast user switching': leaving the session and the applications open and active till the user gains access again.

Integration of nice things such as databases to make life easier: automatic mapping of city/zip codes and such.

But I guess I'll keep on dreaming :-)

Chris Donohue 11/19/03 08:36:49 AM EST

OK this may seem like a really silly comment but one thing that has always bugged people when I show them linux is floppys. Yes you can setup automount to use a floppy. But there's no autounmount. Users have been trained to wait till that green light stops flashing, then take their floppy out. Is there a way to change this behaviour so it works more like other OS's in this regard.
-cpd

Linux Newbie 11/19/03 08:29:39 AM EST

Even I know you shouldn't run as SU all the time. There's absolutely no reason for it. Anything that needs to be run as root will prompt you, or simply type su at a prompt.

I have to wonder why Oliver is trying out Linux in the first place. It sounds like he wants everything to run like Windows. There's a solution to that - run Windows!

I've installed RH 9, Fedora, and Mandrake 9.2. I tried out the live eval of Suse 9. I'd have to say that Suse was the easiest, with Mandrake a close 2nd. For other newbies, I'd say give Fedora a miss for now. What I liked about Suse was that I didn't have to go around and install things like Real Player, Flash, etc. Ya, I know it's proprietary stuff, but it was just nice 'cause I'm still new at this and it's not quite second nature yet. Plus, if I'm going to recommend a distro for non-techy friends, it has to be as easy as Suse.

So if Oliver is going to try Linux again, try the live eval for Suse 9. I was watching news with Real Player over my dsl connection within 5 minutes. I'm not sure how it could be much easier.

I haven't tried Debian - but I think I might give it a go for apt. I really liked Synaptic for RH9, and I haven't found something like that Mandrake yet. I'm guessing something similiar is out there, I just haven't found it yet.

kaptain kernel 11/19/03 08:03:18 AM EST

Oliver Barthelemy said:
"installs usually don't just WORK. And, I can wiggle my way out of Windows's install issues, not Linux's. I tried Mandrake, Red Hat, Debian."
Oliver is either lying or just doesn't know what he's doing .
Mandrake 9.x has probably the easiest install I've ever seen - same goes for SuSe. In fact each were EASIER than Windows to get going.

Kapital (personal finance manager):
http://www.thekompany.com/products/kapital/

Access clones:
Rekall (recently GPL'ed):
http://www.thekompany.com/products/rekall/

Kexi - currently being developed
http://www.koffice.org/kexi/

Screenshots of Kexi look REALLY good - i can't wait for it:
http://www.koffice.org/kexi/screenshots.php

Oliver Barthelemy 11/19/03 07:18:39 AM EST

Improve that OOBE !

I've tried to install Linux on one of my PCs 6 times in the past 2 years, and always ran away because:
- installs usually don't just WORK. And, I can wiggle my way out of Windows's install issues, not Linux's. I tried Mandrake, Red Hat, Debian. Best so far is Knoppix, with 'only' some display bugs. I know install scripts and tests are not gratifying work, but I'm busy. If after one evening (night !) Linux won't run, it goes out the windows ;-)
- Not only is Linux' ergonomics bad, it is also bad in a different way than Windows'. I never figured out how to change my screen resolution, to update drivers, what gets done in xconfig, what in K-config (?)... How to do system things (update drivers...) should either be crystal-clear, or same as Windows.
- settle on standards. I still don't know if I should use Gnome or KDE as a desktop, and each Linux friend I ask pretty much has a different answer, including other options. NO, I don't have time to try them all and make my own choice. Especially since I normally fail to even get one running correctly. And I want my friends/colleagues to have the same config I have, so I can show (-off) stuff, and not look dumb if I have to use their PC.
- don't forget entertainment. I want dumb games, mp3 and Divx stuff, right off the bat.
- super user thingy. If as a normal user I cannot do everything I want (such as installing/updating programms, moving files...), i'm gonna log on as superuser all the time. The user interface should make screwing up my system when I'm an SU a bit hard. Because (repeat) I'm going to be SU all the time.
- directory structure and shell. Another case of "either cristal-clear or Windows-like", only cristal-clear don't exist ;-). I open several DOS shells a day, even now. UserLinux' shell should be DOS-like. And I should be able to find my way around my hard disk, to edit those pesky config files.

I never got there, but I'm told once you do, Linux desktop apps are pretty OK. Their Windows ports are ;-)

Thanks a bundle to the community for all the hard work. You've got the servers... you'll get my PC soon ;-)

Olivier

David Durst 11/19/03 06:33:43 AM EST

Oooh I forgot to mention one MAJOR THING!!!!
When the HELL is linux going to get a version of AOL!!!
Access to the worlds largest dial-up ISP, not to mention
most desktop users still use their service and consider it the "INTERNET"

Yeah that might be nice.

David Durst 11/19/03 06:30:13 AM EST

Glen Schuler I hate to inform you that OS X is poor to say the best about it. Yeah it looks pretty but it does the basics very very bad! for instance after a few months the system slows to a screeching hault because disk frag gets so bad. I have yet to ever have to DEFRAG/FIX a FS under linux since 2.4 Kernel. That and OS X is extremely insecure there is very little seperation between Basic User & Admin and no clear way to define user priv.

Maybe in a world where a system is to be used ONLY in a home and ONLY have one user would OS X be a good system but then again if I wanted that I would just get XP and pay $1000 less for my computer.

Pradeep Sharma 11/19/03 06:19:34 AM EST

I have been on linux and freebsd since last 2 years. I think the click and run should be done the freebsd ports way. That will require an interface and m/c to host the gz files.
Neither mozilla, nor koqueror galeon etc are as cool as IE. Sorry to say this but fact is fact. I always appreciate the good but do have guts to point on the bad. Anything that is written for web is mostly tested on IE and not on mozilla people do but not always and html is not xml ( i hate html). This makes the page look and feel bad.

A normal user just wants a messenger, a very good browser like opera(IE, that doesnt crash, ie do crash ). a good word formatting doc - openoffice is cool. And a nice interface that wraps it all and provide the same look and feel and short cuts (kde).

I sit on windows xp and use putty to linux. Its having kernel 2.4.x and glibc. But would love to sit on linux desktop. I would also give some efforts in developing the stuff. I think UserLinux will only conqueor windows. IF GRANDMA finds it difficult to differenciate what she is using is windows or linux.

PS: Lets make a news group for userlinux.
Cheers.
Pradeep Sharma

Glen Schuler 11/19/03 06:15:52 AM EST

Um, I hate to break up this little party, but someone has already done all of this. Has anyone here heard of Mac OS X? Why don't all stop trying to reinvent the wheel and help push the one that is already rolling?

David Durst 11/19/03 05:55:08 AM EST

Martin Sevior, AbiWord is no where NEAR OO and nor is Gnumeric. Primarily the issue with both is EMBEDDING EACH other for instance putting a Graph in a AbiWord doc and secondarily stability. I constantly check in on both of these projects and compare them with OO. They still do not measure up.

Alex Beels, when is the last time you went to a store and got eithier RH9 or SuSE 9???? These issues with drivers you mention is for the most part NON-EXSISTENT - outside of the minor issue with some video cards (Which I don't believe is really a issue anymore last time I checked) and some onboard RAID-IDE controllers. For the most part I believe in my experience of having to install both Win2k, XP, and RH9. RH9 does a FAR better job at detecting HW that any windows version. Oooh and thats RIGHT I don't need a VENDOR DISK!

Ooh and I believe your first issue has been resolved with MetaCity.

And on your accounting comments there is GNU Cash but I have yet to see that project reach a point in which I would suggest its usage for every day people (Mostly ease of use)

Raymond Wilson I am with you but all of your points on sending email and surfing the web etc.. also come along with a expectations of certain functionalities with in each. This appears to be the underlying issue, not the presences of programs that will do the basic functions you have suggested.

The real underlying issue with what you have suggested is the integration of all of them, that is when I open a Document in open office and click on a URL it needs to open the right browser, when I click on a email link it needs to automagically open the E-Mail program, when I download a MPEG it opens the player and so on and so on.
Which ultimately leads anyone that is going to make a desktop distro to make some hard decisions in which software to dist. because when you have 5 programs that can PLAY DVD people get confused.

So I would suggest to UserLinux first not to start laying out requirements based on some abstract requirements like
"User needs to surf net"
but really find out what that means to the end user.

Secondly identify what applications fit those requirements the best and only distribute those.

Thirdly, INTEGRATE INTEGRATE INTEGRATE (It is kinda of the same issue that Alex mentioned with USER INTERFACE)

Andreas Mohr 11/19/03 05:48:50 AM EST

We should probably introduce package ratings for Debian packages, such as "stability", "usability", "docu rating", "desktop importance" or "desktop ranking" and "GUI rating" (unless such a thing already exists for packages, but I haven't seen it yet).
These are just examples, but in a real implementation it could probably be more fine-grained, more indicators.

This would help solve a big problem with all kinds of GUI package managers: while they do many things to improve package management ease, they still overwhelm people with TOO MANY packages.
People don't want to see all 45 CD players a powerful Linux
system offers - they want 2 or at most 3 that WORK, and work WELL.

By adding package rating indicators, package managers can automatically focus on important packages. And if a user wants to see more packages than the default package rating thresholds allow for, (s)he'd just adjust the thresholds.

Oh, and of course we should also have a specific rating to make sure desktop package managers don't even list utility libraries in most cases - listing should be restricted to important programs only.

Implementing this properly would go a LONG way towards user-friendly package management.

Besides, if a package manager has a "docu rating" threshold of "4 of 10", then this would provide some incentive for package maintainers to include the minimum amount of documentation/help to reach this threshold.

How the actual rating would happen is another question:
who decides the actual rating, and which criteria do we have for that?

IMHO such a mechanism is sorely needed.
This should be introduced in the Debian core architecture ASAP IMHO.

Andreas Mohr, ACX100 wireless driver and Wine developer

Raymond Wilson 11/19/03 04:01:32 AM EST

Many people have supplied useful comments in this feedback forum, yet I can't help thinking that the point has been lost on some!

This discussion is about _User_ Linux, the Linux distro to claim the desktop for the great unwashed millions upon millions of computer users in the world ranging from the early teen doing school homework assignments to mum & dad doing the IRS tax returns to Granda sending email to the grand kids.

These users have a single, simple objective to their computer use: It should just work!

Forget anything that requires even the tiniest bit of computer related knowledge. If they can't achieve a desired operation (eg: Send the email to the grandkids, choose to play solitaire from a list of games, fill in their tax return, write that school essay) in more that one or two clicks to run an obvious program from an obvious location (eg: start menu or even just great big obvious icons on the desktop) then it has already failed!

Software installation should be single click affairs, with no questions asked.

It doesn't have to be the greatest piece of eye candy in the world - it just needs to work, intuitively, consistently, simply and easily.

95+% of the 'great unwashed' computer users will do these things:

1. Send email
2. Surf the net
3. Write small-medium sized documents in a Word type application.
4. Play some simple games (solitaire etc, or web based games from ShockWave etc)
5. Play the odd fancy game like Harry Potter etc (though this will be more the kids rather than adults).
6. Do some simple spreadsheet tasks (but only if they are adventurous)
7. File their tax returns...
8. Play music and perhaps DVDs...

The people who say UserLinux should seamlessly move between networks or that it should have better system configuration tools don't need UserLinux - they already have the know-how to do it using other distros and UserLinux would only frustrate them.

In some ways Knoppix has a good model: You put the CD in the CD-ROM and reboot. A minute or two later you are looking at a Linux desktop. In most cases you don't even need to set any of the startup parameters - IT JUST WORKS. Typically the only thing you have to tell it is the phone number of your ISP and your username/password. That's it, and the great unwashed can cope with it. Much more and they are lost and UserLinux will have missed its mark. Of course, Knoppix then provides an enourmous set of menus containing names of programs that may as well be heiroglyphics to these people :-)

If you make UserLinux capable of these things then it will go far and as the computer illiterate of the world start to use it and become familier with it they may wish to explore further. But until then it needs to be the easiest distro for the person with few computer skills to use.

Cheers,

Raymond.

tucker 11/19/03 03:42:55 AM EST

i think the biggest problems i ever had, as a newbie were dealing with APM and compiling my kernel. to make linux a non geek system, you'd need to make the kernel completely invisible to the standard user. this ties in with the whole driver/hardware detection/upgrade issue. your average computer user can barely figure out how to install drivers for their new video card or printer, let alone know whether they need video4linus or what type of USB hardware support to add in. even then, compiling the kernel is going to be very confusing for a layman.

as stated before, installing linux on a laptop and getting power management working properly is a pain. i've all but given up on my laptop being able to go into suspend mode and it's a tuxtop.

Rudolph Froger 11/19/03 03:38:30 AM EST

A lot of non-professional users, do not have their computer running all day. When they want to use their computer is should have a very short start-up time. Linux distro's for desktop users are lacking this.
I switched my parents to Linux two weeks ago. Their first (and only) complain was the start-up time of Linux and OpenOffice.

Alan Chandler 11/19/03 03:11:23 AM EST

Bruce asked the question - what do people think about Konqueror. I use it all the time because there are two minor little items in konqueror that I haven't found in all other browsers. 1) An up button, that goes up one level (directory) of the URL. by stripping the tail off of it. 2) A button to clear the URL area, so I can centre paste a new URL into it

Sea Dragon 11/19/03 02:54:59 AM EST

So many great idea, it is great.

Well, for my view,
- About Application in Desktop
Let people have choice to choose their favor. Build up Application server that show their needs and allow they choice. Everyone will happy to find they need
Think : how to attract more developer to create Application for Linux
how to let application to work with different version kernel

- About OS System configure
For normal users, must be easy to maintain
For company system administrator , let them have advance configure on Desktop. ** easy for them to maintain large number of Linux Desktop **

Life is fun and happy and no angry

Alex Beels 11/19/03 02:17:22 AM EST

A couple of comments:

jason - At least as far as the network is concerned, you can already do what you want with redhat-config-network. I take my laptop back and forth between 5 different networks all the time, and I can switch between them with just a couple of clicks.

Martin - People who suggest OpenOffice are not just jumping on the bandwagon. Though it is slow and clumsy, there are a lot of uses for which OpenOffice is the only reasonable replacement for the M$ product. The feature set is 'not much different', but for the people who are sitting on the fence, those few differences are often critical. I'm writing a dissertation which uses three languages, one of which is written vertically, and I have to exchange my documents with hard-core M$ serfs all the time. I can just barely do that with OpenOffice, but not yet with Abiword. I sure wish I could use Abiword or KWord for my work. Maybe next year.

I'm not trolling, just pointing out that a lot of us *did* do hours and hours of homework, even if we came to different conclusions than you.

Martin Sevior 11/19/03 01:58:47 AM EST

I wish people who want community involvement in Office Suites would look beyond Open Office. Gnumeric is already a better program (in oh so many ways) than the Open Office spreadsheet program and AbiWord will soon surpass OOo writer. Both programs are an order of magnitude easy to develop on. eg AbiWord has around 300,000 LOC (not counting the duplicated functionality in the different Front Ends) as opposed to OpenWriters 3 million for a feature set not much different. I also personally believe that AbiWord is a far more usable program than OO writer and much better suited Gradma already.

We have ways of making a presentation program from the AbiWord codebase which would not be too hard and could easily to be extended by community support. Finally GnomeOffice has a great database application which is a feature which is totally missing from the OOo framework.

*Please* people who have grandiose ideas, *Do some homework* instead of jumping on the latest bandwagon.

I'm not sure of the status of the KOffice project so I can't speak of them

There is a good reason OpenOffice has very little community input.

jason 11/19/03 01:54:46 AM EST

I'd like to be able to change network environments without thinking about it. I bring my laptop to classes at the local college where I have to use DHCP and would like to be able to find (and possibly cache findings for next time) the local printers, (replay this point several times) then I visit a friend's house or go home where a bunch of fixed IP addresses are shared and where there are different DNS, printers, gateways, etc, and I might have to use my laptop for IP forwarding over dial-up or broadband.

Sometimes I also have particular mindsets for my laptop:
- sometimes I'm just spending time so I just want a browser and email client running
- sometimes I'm doing database / web development so I want my database, web server, an IDE and possibly an xterm running
- depending on where I am and for how long, I might want my firewall up

Setting up my machine to be familiar with different network setups and usage profiles for a single user should be easily accomplished through an intuitive GUI, and switching between them should take 1 or 2 clicks.

Also, as a non-root "power user", I should be able to mount shares, install applications, publish one of my directories to the LAN, and other things without being given the power to destroy my install or a partition table. Basically, a "nearly" root user.

Alex Beels 11/19/03 01:50:47 AM EST

Some of the ideas in these comments are great. Unfortunately, the best ideas are also impractical in the near future. They describe an ideal environment, not the next generation of Linux.

I think the focus on applications is a little off in this discussion. The biggest obstacle I have found for people trying to adapt to Linux is consistent operability. This falls into four categories:

1) User interface. Xaw + QT + GTK + Tcl/Tk + WxWindows + SOL (OO.o) + Motif = chaos. All widgets in all widgets sets must look identical and act identically. Fonts, standard shortcuts, tab navigation, mouse behavior, and drag and drop must be identical in all applications and all environments. (I.e. get *rid* of middle-button cutbuffer pasting, no matter how much we all love it. We can only have one cut-and-paste paradigm.) The user should be completely unaware that there are multiple widget sets and multiple desktop environments. KDE? Gnome? The user shouldn't be able to tell the difference. Let programmers have all the different tools they want, but hide the diversity from the user.

I cannot tell you how important this is to getting users to accept the environment. If I sit someone down in front of a 100% KDE machine, they usually say "Hey this pretty cool. Too bad it doesn't have Quickbooks." That is a surmountable objection. We can write apps. If I sit someone down in front of a KDE/Gnome/Motif machine, they invariably say "Huh? Why doesn't anything work?" This is a deal-killer.

2) Configuration. This has been well covered already in this discussion. (A) A hands-off install with hardware discovery, default settings for *everything* and dependency resolution. We're almost there with anaconda/kudzu/apt/yum. (B) ONE set of GUI configuration tools that cover EVERYTHING. Nothing missing. No duplication. Located on an obvious menu. Again, we're getting close here, except for little bugs and the duplication issue.

3) API. Library API's should only change on major version changes, and then only to add new calls, not to change or eliminate existing calls. Why is Windows such a successful platform? One big reason is that the API hasn't changed significantly since 1995 and still supports applications from way before that. Inconsistent API's are the reason why dependencies are such a pain on Linux. If you didn't tweak your API just a little bit on every point release of your great widget/networking/what-have-you library, the Debian testing tree wouldn't be filled with applications which don't quite work and we could kiss specific version dependencies goodbye. Also, right now no proprietary software vendor can write software for Linux with any guarantee that it will still work in six months or indeed that it will work on any distribution other than the one they built it on, unless they do a huge static build. Linux has to supply API's as stable as the core Windows DLL's before someone can write a piece of proprietary software and know it will still work ten years later. (And we unfortunately *need* proprietary software. Some software can't be written by anyone other than a big software firm. Tax packages are a good example: they need to be aggressively updated all the time with the cooperation of a big team of lawyers and accountants. You can pay that staff on a non-proprietary model?)

A stable API is actually really easy to achieve. All it requires is some discipline from the programmers.

4) Drivers. There isn't much we can do about driver availability per se, except to continue to lobby hardware manufacturers. However, we can do something about the relationship between drivers and the kernel. (Note that I am not a kernel hacker at all, so what I am describing below might not be possible. But my suggestions are not technical, they are look-and-feel issues based on my experience as a computer user and administrator.) The kernel compile procedure must be completely revised, and here I realize I am talking about something quite beyond the "next generation of Linux." Simply put, there should never be a reason for the user to recompile the kernel except for a major upgrade of the entire operating system. I *think* what this means, at a minimum, is that all drivers should be available only as modules, and modules should be developed and distributed separately from the kernel. I.e. short of a major version upgrade, I should be able to swap in different versions of the kernel or different versions of some module without the one breaking the other. The kernel provides a stable API for the drivers and the drivers write to the API. The internals of each should be hidden. Maybe it also means that the kernel needs a plugin-like architecture for dealing with subsystems, so that USB support could have been added to an old kernel without recompiling or upgrading. Maybe what I just said above was "Linux should be the Hurd?" Maybe the kernel already has these abilities but the tools and distribution methods don't exploit them. I don't know. But, as long as hardware issues cannot be fixed simply by downloading a driver and maybe rebooting, Linux will be unusable for an average home user who doesn't have a live-in guru. How often does a Windows user touch the Windows 9x kernel? Never, right? That is about as often as a regular user can deal with. Linux kernels have to be able to support the same pradigm of "Yeah, the system on my computer is 5 years old and it sucks, but all I do is download some new drivers, and my new post-firewire-super-terabit-mp3-jukebox runs just fine."

Once we have these basic operability issues settled, the apps will come easily. In fact, except for a graphic database client and accounting software, we've already got most of what we need. I gave away my Windows box and live just fine with KDE/OpenOffice. No WINE, no nothing. But then, I am not a small business and I am not a major media/game consumer.

Fun 11/19/03 01:27:16 AM EST

........

If you want Linux Desktip to have future, USerLinux must pay good attention on the children. Let them to find funny in UserLinx Desktop. PS2 Game....smoe other great game. Children are potiential Linux Client in future.

Well, it is not easy. I think UserLinux may not do these all things because of the limited resources.

My children like Game, such as PS2. Now M$ have too many game in front of Desktop and have more power on Game society.

I just wonder what my children really want to select "Linux vs M$". It is so interesting.

Fun 11/19/03 01:26:46 AM EST

It is great that everyone have good idea to Linux Desktop.
Well, for my additional point of view,

Have question for Bruce Perens :
- What is the target users for UserLinux's Desktop?
If for IT professional, I think there are too many Linux Desktop exist now, LinuxUser just based on them for some modification. Then good enough.
If for coporate user only, UserLinux just add in some Account package, good office app, virusless. That is good enough. Don't need do fancy because there are too many IT professional in the coporation. They have money.
If for unskill users, LinuxUser must have great change to Desktop Linux. Let people stay in mind that when they buy pc in the shop and ask "Do you have Linux PC for sell. I look for one". They ***will not*** ask that "Any update window PC such as Window 2003 for sell, I look for one"

David Durst 11/19/03 01:11:45 AM EST

I am suprised some of the obvious PITAs of running linux on the desktop are note addressed in this article. All I hear is the same old same old which never addresses the needs of a corporate desktop. First of yeah Open Office and Evolution are still lacking some of the nice features.

But the real issue barring linux from entry onto the desktop is Simple Security Control administration for example Windows 2000 has defaults settings - "User can install software" etc.. etc..

Secondly PLUGINS!!!!! I keep saying this but I have not seen a distro address this well besides SuSE 9. Currently I admin a network of 75 linux desktops and my biggest complain t is that Plugins don't work - for example CAD,Java, Realplayer etc...

Thirdly is addressing the issue of Web Site lock out, if a Linux desktop distrobution is to work, any browser needs to be able to FAKE being IE 6 or whatever the latest version is or many sites lock them out - I am not talk about the CNNs of the world but for instance ADX EDI Internet exchange locks out users that do not use IE, this seems to be a continuing trend from my vantage point - WHO WILL A DESKTOP DISTRO ADDRESS THIS ISSUE?????

Kevin C. Redden 11/19/03 12:57:27 AM EST

I 100% agree with the installing of software in linux. That's one of the major walls, I'm running into. Lately I tried to use apt4rpm (aptget), and again I was frustrated. Instead of having an util that I could type 'aptget opera' and it'll install Opera, instead I get 'can't find that filename.'

What I found though much digging, is that in essence aptget is only set up to upgrade, and update your system. To get what I posted above, I have to do a lot of undocumented, and unsupported stuff, that frankly should either be IN aptget, or else set up the system itself to do it's job.

I like the idea of having one download directory where it looks. Would save a lot of problems trying to find where the file was stuck when downloaded. But It should be a standard 'mkdir' type directory. Not a special (propritory?) type directory which is how aptget seems to work. I haven't managed to get this working yet, so I'm still sort of in the dark.

I'm not one that only uses the GUI, and damm the command line, but I need a program that *works*. Aptget doesn't work yet.

Hope that's proven helpful. I'm really wanting to get into linux, especially since MS's talking like they're going to take control over a computer I buy, to please their big company friends. Hence either I stay in XP Pro for ever, or else I get into an OS *I* control.

I'm >< this close to tossing XP Pro.

Kevin

John Austin 11/19/03 12:51:44 AM EST

Take the "shared costs" money and buy out Lindows/Xandros/Lycoris then distribute as necessary. These are all close to what you and others want and two of them are debian based. All three are unlikey to survive, at least one will likely merge. Baring the above I have the following ideas:
** stability, integration, consistency, simplicity and easy of use out way all needs for individual applications. -- Its the System not the apps. Don't misunderstand, apps are essential, but you already have interested 'investors', so you don't need to attract customers so much as delight them. Furthermore, not everything in the software world 'needs to be free' but the OS ought to be. Although desktop app. companies have not found the linux market lucrative up to this point, its a matter of market share more than 'linux mentality of not wanting to pay.' Oracle is making plenty of money on their linux apps and when their's a market, Quicken will follow.
** for the enterprise desktop, zero admin costs are essential.
** For the larger market - beyond the enterprise desktop, you need to offer something more than just a cheaper alternative to microsoft. Killer apps have been the hallmark of the computing revolution - Visicalc, MacWrite, Pagemaker, Photoshop, Mosaic-Netscape, WinAmp, AIM, iMovie. Are there any future killer apps gestating in the open source and/or linux world?
** short of killer apps, there are several killer system features that linux can offer that no one else does:
***CD demo - this is essential to intorducing someone to a new OS. the one problem with the current round is that you can't demo and install. Knoppix is primarily a CD linux (it can be installed, but this is not a newbie thing). Meanwhile SUSE demo is just that an independant disk. I want a disk that I can demo, then hit an install button and I get the exact same thing I just saw permanently installed, with no-brainer partitioning.
*** the second killer system feature is fast boot. Although currently not for the faint of heart, perhaps the parallel load techniques described on IBM's site earlier this fall could be perfected and incorportated into a distribution. This would allow linux to boot much faster than windows on the same hardware.
*** finally, borrowing from the PDA and embedded linux worlds, a really good instant on/sleep combined with linux stability would give it the snappyness of a PDA and the reliabily of the good-ole-days when you left your unix workstation powered 24-7. You could come back on Monday morning to your desktop with the edit session and windows just the way you left them on Friday evening.

Chris Bruner 11/19/03 12:41:31 AM EST

I wouldn't limit UL to aptget. Take a close look at Gentoo and where they are going. Some very exciting things are developing in the package management area. (prolog expert system to handle dependancies). I used Debian for a while, switched to Red Hat,Mandrake and landed on Gentoo. It's now my favourite, and it's getting better all the time. Package managment is currently great and is getting better.

Power managment would be nice. (thinking about my own old box).
Why not add features that you can't normally get in windows.

Speech systhisis (using festival). I've got mine set up to tell me when I get mail from important people, and to read me the subject line for some people. A nice gui to handle setting this type of thing up would not only pass the grandma test, but jump the usefulness up a notch.

Voice dictation built in. (I'm sure there is a package out there (IBM, come on make Viavoice open source :) It's not something that I use, but my dad (the grandpa test) uses it regularly (and likes it).

Schedulers (so the notebook by the phone can be used for notes). Why not have a list of friends so your scheduler can coordinate with their scheduler to make appointments. (Dentists, vets etc would also benifit). The idea is you plug in the days/times you are available, and where you want to go(as well as best choice). It sends email to the other party, who's schedular automagically takes the email (special tag in the header) and confirms one of the days, and mails back replies. Your scheduler gets the reply and updates your schedule. All this happens in the background.

Just a few ideas.

Lodragan Draoidh 11/19/03 12:34:20 AM EST

One of the major complaints I hear is about differing interpretations of the file system hierarchy [pathname.com]. While I think standardization is good, I also believe developers should have a certain amount of flexibility - which the standard allows. The key here, I think, is for the distributions to honor the locations that the developers established for their files - so compatability crosses all boundaries and documentation can be maintained by the developer on the particulars of his application - instead of the distributor. In cases where the application creates problems due to inappropriate placement - the issue needs to be raised to the developer to correct his implementation; distributors would have the option not to include the application/system if it was too disruptive - but that is all (more than this and the distributor can cause more problems than he intends to fix). Developers need to understand the standard; distribution creators need to cede the responsibility for application locations to the developers - with the right to veto bad locations from entering their distro until corrections are made by the developer. This way, no matter which distribution you are using, foo.ini is located in the same place every time.Related to this, and probably more frustrating for end users, is when application developers make assumptions about libraries and other applications that exist on the system during the build. For hard core *nix system administrators this is no big issue - something they have been dealing with for years; however, for a general purpose workstation this has to be idiot proof. Coupled with standard locations includes being able to check those locations for particular files, and if not found, have the confidence to load them for the user, rather than simply complaining and dropping back to the command line. Again, the onus is on the developer to include all parts necessary to work with his tool (perhaps even going so far as loading a different library in an alternate location [sub directory in standardized path location] - then changing an environmental variable used exclusively by the application to locate it without disturbing an existing library or any applications that depend upon it - lets definitely do it smarter than Microsoft DLL hell)These two items coupled together would make installation and maintenance across all distributions easy - and dependent on the documentation and careful work of the developer community - instead of left at the whim of the distribution agents - who are not on the same sheet of music. If developer X creates app Y and puts it in location Z - then Z should be where everyone finds Y when they look on their system.Finally, I think easy to use tools for administering very clearly standardized core items (the rc.d run level scripts, crontab management, X configurations etc...) should leverage existing text based configuration files. Lets not get into the trap of reimplementing the Microsoft registry - as a single point of failure. Up to this point these types of tools have been adhoc; someone needs to take the ball and run with it to create something that is clearly superior and usable for all distributions that intend to target the niave user (hmmm - sounds like a good open source project to me - maybe a python Tk gui with a builtin command language parser for power users... :)These are the core items I think are critical to a successful linux desktop to compete with Microsoft's dynasty.One additional frill I would suggest:Implementation of a better 'Annotea' [w3.org] W3C implementation, as seen in the W3C Amaya [w3.org] web browser, within more full featured web browsers (in a nutshell allows an end user to make and view their own annotations of documents [meta data] on the web - without altering the documents themselves - or having permission to for that matter; users can keep private notes or share notes on servers. Links are viewed on the webpage itself; if you have not tried this I would download it and give it a whirl. This concept is a killer app in my mind - but its a crude testbed right now).

roly 11/19/03 12:16:05 AM EST

I agree with most of the comments above. Especially the hardware detection, hot plugging usb etc. I only keep one windows box for my tax preperation.

What I have often thought is we need to setup a central open source bank that we could all contrbute what we could afford that then could be used to fund the developement of projects that do not have a mass appeal such as tax preperation or the smoothing of the users experience with software installs etc.

David Jeske 11/19/03 12:01:52 AM EST

This approach is naive. Desktop Linux will never compete with Windows if the goal is to replicate all applications as open source. Desktop Linux should be focused on bringing existing 3rd party software developers to the Linux platform.

Jason C 11/18/03 11:36:33 PM EST

Oops...totally forgot:

Make GCC much, much, much, much better. Best of breed IDE with best of breed optimizations for every chip that has distro support. Must be able to pass Intels compiler for quality of code (this alone could be the most difficult thing to do. Come on IBM, we need a major player to bankroll this part and give the code away WITH NO STRINGS).

Jason C 11/18/03 11:32:13 PM EST

1.) Standards. Brutally enforced look, feel, behavior standards (think OS X)

2.) Ditch X windows. It is absolutely unneccesary for a workstation. Video needs to be much tighter bound to the hardware (well, audio too). Ideally 100% of video rendering should be done on the VPU not the CPU. (think Longhorn)

3.) Obfuscation of the system to the end user. If the end user so chooses, there should be zero interaction with the nasty low level system at all. It should all be point, click, just work.

4.) Never patch or transparent patching. The biggest gripe of any system is that you have to mess with it. A good system should never have to be readjusted. Bruce should be aware of this concept from dealing with embedded systems.

5.) Distribute binaries. Precompile as much as is humanly possible and then on-the-fly compile the platform specific code without any user interaction. All programs should come with every single piece necessary to make it run. Everything runs out of its own directory in a true sandboxed environment.

6.) Distros should be light and fast. I don't want to see a single server daemon on a workstation. Daemon are applications and have nothing to do with a basic OS. They can be packaged and managed seperately.

Wish list:
Instant on

Kernel lives in firmware

SSI clustering with full DSM and migratable sockets (and threads) that is transparent and in the kernel

Supports every piece of hardware known to man

A thin virtualization layer (a la VMWare?) for stability and predictability

For my hair to grow back...;-)

Jeremy Petzold 11/18/03 11:31:54 PM EST

as far as package management, I think it should be even M?ORE intelligent than what is available in apt.

the user should be able to download a package from a site, double click it, and when the package management system is checking dependancies, if any are missing, it should go out to an apt source (or use auto-package) to get the dependancies, if the dependancies are on some weird server because they are not placed into a distro source, the package can have a header in it that lists the sources to get dependancies, pass them to the package manager, and go to those locations rather than an apt source (which would be default in the event of no package headers with sources for dep resolution). this would eliminate the need to go to the command line for the user, and eliminate the need for a front end to a command line tool other than some prompts to the user to authorize the download.

when Linux package management is as easy as that, then it will be truly easy.