Talk:Proposed Freekbox Changes

Argument for Ubuntu

Michael:

I saw your email to the Adoption-teachers list and thought I would immediately jump in here and throw in my thoughts.

Firstly, my recommendation would be to go with Ubuntu, all things considered but I want to go into why I say that in a little more depth.

I've been teaching the adoption class for about a year now (maybe a little longer) and one of the things that I see is we get a wide cross-section of user-students in the program. That said, we also see a lot of folks (particularly older people) who are toward the lower end as far as computer literacy and it is this group of folks (the computer illiterate or barely literate) people I'm thinking about. These are folks who, for the most part, don't want to be geeks. What they want, the reason they came to Freegeek in the first place, is to earn a computer that they can take home and use. They want to plug it in and have things *work*, more or less automagically. Now, most everyone reading this is going to know that it would be very difficult to give the users' that but I believe we can endeavor to make things *easier* for the user.

Debian is a great hacker's distro and that is precisely the weakness of it for our program. The folks who get the freekbox aren't necessarily hacker's and they don't *want* to be hackers. As much as that may be something that seems alien, it is still the truth that folks want to get their freekbox and take it home and then have a Windows-like experience without the Windows. If they pick-up a thumb drive, they want to plug it in and have it work. If they have a digital camera or an iPod they want to plug it in and have the magic happen and it works. They *might* be willing to do a little bit of twiddling and tweaking, I'm not saying that folks are looking for miracles but for your average end-user, the idea of hacking at, for instance, smb.conf does not equal crazy delicious fun. Now, for me, hacking smb.conf is a perfectly fine way to spend an evening but it's not for most of the folks who I have had the pleasure to teach in the last year.

We get them for three hours, folks. That's it. We can change the education model a bit and I'm hoping to work with Liane and Michael on expanding classes but the only class we can reasonably assure that they'll be at is the Adoption class and that means three hours and a bit of change. In that time, we need to give them enough to be able to take this machine home and do *enough* useful with it that they don't just get it home and treat it like a doorstop that can play mp3s.

Of the distros listed, I think that Ubuntu would be the way to go. Although I'm a SuSE bigot, I've installed Ubuntu and Kubuntu and played around with them on my desktop machine (and even had Kubuntu running on my laptop for about a week before I decided to just use SuSE and be done with it) and I'm very impressed from the point of view of looking at it from a teacher's eye. (And also from a hacker's eye but that's a different story.) The folks at Ubuntu seem to 'get it' that if Linux is ever going to move solidly into being an end-user operating system, it needs to be usable. Hardware recognition is a big deal for folks and users shouldn't have to huck their freekbox down to Freegeek when they buy that spiffy CD-RW from the Freegeek Store.

I would suggest staying with the KDE desktop environment. It's powerful and still basically very friendly to use. I also think it's more intelligently designed (again from an end-user perspective) than Gnome although I know that a lot of folks would disagre with me on this.

cheers Aj Davis

Ladyfractal

From this (and Michael's email) I get a few concrete tests we could take away:

1. thumb drive

   should be the same as the mass storage item - MW

2. digital camera
3. iPod

   anyone have one we can play with? - MW

4. automatic detection and mounting of mass storage devices

   works with both ubuntu and default sarge, but is only reliable in the gnome desktop

The main things that I am looking for in our next Distro are

1. Automagic plugin Device use
2. No user should have to use a terminal ( I know it's strange, but some people hate it )
3. Automatic updates: Sure Debian may have the quickest bugfixes, but how many adopters have NEVER done an update because they forgot all about apt/synaptic
4. Community: While the Debian Community is geared more towards linux Admins, the Ubuntu community has a very strong leaning to Desktop Users and Linux Newbies. This includes forums and documentation It would be great if the Free Geek community could plug into to Ubuntu community.

Dave 15:09, 15 Apr 2006 (PDT)

We should create a prototype of each of the three options and see how each performs on these tests. Maybe there are more things people want to try out. If so we can add tests. What I read in Michael's email, however, is that while our version of debian (the FreekBox) has some problems and Ubuntu has the advantage over it, it seems that a standard Debian install has also solved some of these problems. It'd be nice to know that we're jumping distros based on quantifiable differences rather than any fuzzy feeling any of us may have.

RfS 17:15, 11 Apr 2006 (PDT)

One argument for Debian

Sooner or later we'll grab a FreekBox off the shelf and throw it into place as an infrastructure box. When we do this, we could have a mixed network environment at Free Geek, the standard debian boxen we use on our servers, and an Ubuntu box with its different security model. This is less maintainable. Also not a very big issue, because we could have a standby debian system or two ready to go stashed somewhere awaiting this eventuality. RfS 17:15, 11 Apr 2006 (PDT)

We have a network Debian Installer on the build network, so installing Debian on a box is as easy as Ubuntu. Should be no problems there. Dave 14:57, 15 Apr 2006 (PDT)

My point is that when the front desk computer fails (for example) we grab a FreekBox (already installed) and drop it into place. We don't grab a box and install before putting it into place, because there's not time. Then what happens is the box ends up staying on the network. Of course, this is a small issue because "we could have a standby debian system or two ready to go stashed somewhere". RfS 08:29, 17 Apr 2006 (PDT)

Richard- though I can see how the current decision would impact how such a problem is handled in the future, it seems that a suitable "emergency plan" could be easily developed under any scenario. I think this issue should be left out of the decision of what distro is chosen, and revisited once a decision is made. -Pete 11:05, 18 Apr 2006 (PDT)

Argument for SUSE Linux and Puppy Linux

SUSE Linux, http://www.novell.com/products/suselinux/ , is probably the easiest to use and you can buy it with manuals and support at your local computer store for around $59.95. SUSE Linux 10.1 is due April 2006. Make it easy on your customers and give them SUSE. For older computers, flash drives, etc. use Puppy Linux, http://www.PuppyLinux.com . Also see http://www.FreeGeekTacoma.org

Apples and Oranges

It is important to remember that we are not talking here about keeping the current freekbox3 but making sure that what we use fits our needs, our users needs. For example, the default Debian box has a networking tool with a check box to enable windows networking. Most of the administration can be accessed with graphical tools, just as in Ubuntu. In fact most of them appear identical to their counterparts in Ubuntu. I have a box set up in my office which people are welcome to try out. If something seems missing, I will see if it can be easily added, or if it is a real deficiency. I am willing to go along with whatever we decide, but lets do some real comparisons first, not remain stuck with our current concept of Freekbox3 versus Ubuntu. -- MW 17:38, 11 Apr 2006 (PDT)

My main wish in the immediate future is to get a clean list of the features and programs we need, from adoption, store, grant and infrastructure perspectives so that we can do a good side by side comparison. -- MW 17:42, 11 Apr 2006 (PDT)

i feel the User personae process can be very useful, here. Vagrant 17:16, 14 Apr 2006 (PDT)

Functional spec?

Michael:

Do we have a functional specification that I could look at? If I'm going to be at all involved in this process (and I would like to be) and if I'm going to be fair-minded (which I hope to be) then I need some kind of benchmark to work with so that I can be empirical. I understand eschewing the idea of jumping distros either just to jump them or because of some ill-defined criteria such as user-experience.

Cheers Aj

That is what we should be creating here, clear requirements that can be evaluated, such as Dave's "The user should never have to open a terminal". -- MW

Dapper Drake

If Ubuntu is in the mix, we should keep in mind that we are weeks away (June) from the next release. The next release is supposed to be an Enterprise release (I forget exactly how shuttleworth defines that, but it's intended to be some combination of stabler and more feature-rich, and have a longer support period, than Breezy or Hoary were.)

I believe the Drake has Firefox 1.5 also, which is a major improvement in my view.

Is anybody looking at the pre-release versions of this Dapper Drake?

-Pete 15:30, 12 Apr 2006 (PDT)

I have been keeping up almost daily. I have a working prototype in build that needs less and less tweaking with every daily update. Out of the box lucent/agere winmodem support is working, and PCTel winmodem support is on its way. I would like to start putting it on Store Boxes when Dapper goes Beta (april 20) so I can get lots of feedback from builders without putting our adopters at risk.

Dave 15:15, 15 Apr 2006 (PDT)

I'd be hesitant to base anything on a pre-release. We need stability. Sounds like several of the FB3 problems are partly there because we chose an then unstable release of debian to base it on.

Ubuntu's more rapid development schedule will likely mean a more up to date set of software, but might come at a cost of instability.

RfS 16:05, 12 Apr 2006 (PDT)

I think you're missing my point. Certainly, we don't want to DEPLOY a pre-release. But this particular release is a big focus of Ubuntu, with goals similar to ours: stability, long-term support.

The reason Ubuntu delayed the release 6 weeks (till June) is specifically because of the desire to release a final version that is more polished than previous Ubuntu releases.

From both a security and a usability standpoint, I feel strongly that Firefox 1.0.7 and earlier really oughtta be avoided.

Could you be more specific about usability issues? security-wise, debian backports most if not all security updates for firefox. ubuntu had major security flaws in firefox (i.e. "hijack your browser and make you see whatever we want" level security holes) months after they were fixed in debian. Vagrant 17:29, 13 Apr 2006 (PDT)

How soon are we looking to make a decision? If we need to decide before June, that would rule out Dapper Drake.

And, if there are FB3 issues that resulted from testing pre-release software, what were they? Do the same conditions apply here?

-Pete 17:02, 13 Apr 2006 (PDT)

One problem in the freekbox3 is the lack of automounting for USB drives and other added drives. When sarge was in testing, no automounting system really worked well, so we used autofs, which we had used in the freekbox2, but which requires additional configuration for any added drives. The current version of sarge seems to have a very good clean automounting system available, essentially the same as in Ubuntu. The same problem is unlikely to recur, but the problem with the second guessing these things is that it is hard to predict what will shake out in the transition from a testing to a stable release. Also, I very much agree with Vagrant's comment. We can use backports.org for programs like firefox MW 22:06, 13 Apr 2006 (PDT)

Thanks for the specifics. Vagrant, as for Firefox security: I did not realize that Debian was better about backporting fixes for Firefox. I'm curious whether this means that security fixes are made sometimes without incrementing the version number; if so, many of my prior comments are misinformed. At any rate, Ubuntu Breezy is unlikely to ever have a backported version of Firefox more recent than 1.0.7. Here is a discussion that explains that. Here is a link to the list of Firefox security issues.

debian typically backports the code required to fix the security bug into an older version of the source code, to minimize introducing incompatibilities with the previous stable release: debian security faq. this is very different than using a backported package, such as those from http://backports.org, which i would recommend against if at all possible, because they change too often. Vagrant 17:07, 14 Apr 2006 (PDT)

As if on cue, Firefox 1.0.8 was released in the last couple days in Ubuntu/Breezy repositories. I believe this addresses the "security" elements of my concerns with pre-1.5 Firefox. -Pete 10:18, 21 Apr 2006 (PDT)

As for Firefox usability, 1.5 has improved memory handling (very noticeable when using multiple tabs), more intuitive treatment of live bookmarks, and (I think most important) a self-updating feature. Meaning that you don't have to wait for Debian/Ubuntu to add security or other updates to its repository. Also, a growing number of useful extensions require 1.5 or later.

I don't exactly understand how the USB problem results from testing a prerelease, but at this point I'm happy to take your word for it. If Dapper Drake is off the table, I'd say using a Debian-based distribution is the better bet.

-Pete 10:10, 14 Apr 2006 (PDT)

the technology for automounting at the time was not yet functional, or more likely, we just didn't even know better alternatives existed. Vagrant 17:07, 14 Apr 2006 (PDT)

John (blackhole-laptop land)

When it comes to picking any distro of Linux over another, remember our :customer's are not hackers, they don't want all the newest bells and whisles. :There is many good versions of linux out there and i have tried a few to say :the least. Now keeping up our own version takes time and talented individuals, :where as using whats avialbe frees up sstaff time to do other needed tasks :for the needy. Ubuntu KDE Kubuntu Dapper build is very complete and usable, most the hardware expectations of newbies are there. Now downloading good copy to burn sometimes is a pain, but nonless it is good. Now PSLinux is good, but a difficult install, and Mepis crosses to many boundies in open source, but is very easy to install and configure. So the Dapper version of Kubuntu is my overall vote for this week. LOL it is always changing for the better with each new release cycle of linux distro's. ;-)

Praywaror2 17:01, 17 Apr 2006 (PDT)

meta discussion

I asked an important question before, that has gone unanswered for some time. When are we trying to release a replacement/update to FreekBox 3? The (ongoing) discussion of whether or not Dapper Drake is in play relies heavily on the answer to this question. Without a target date, any assertions or denials regarding the testing of prerelease software are pretty meaningless. -Pete 10:22, 21 Apr 2006 (PDT)

wiki is a very difficult way to have a discussion. Vagrant 17:11, 14 Apr 2006 (PDT)

Agreed. I would support an effort to summarize this discussion and move it to an email list. -Pete 10:22, 21 Apr 2006 (PDT)

I see some problems as well. Here are some things that help if we're going to continue in this mode:

• Indent your replies to indicate a thread. (Use : at the beginning to indent once, :: to indent twice, etc.)
• Sign your comments with 4 quiggles ~~~~. This turns into your username and a date stamp when you save.
• Preview your changes to make sure they look right before saving.
• If your comment is going to split up someone else's comment, copy their signature and put it on the end of the first and second half of their comments, so that we know who said what. This makes it a little more likely that when someone says something like "you are missing my point." we will know who "you" and "my" refer to. In fact, try not to split up people's comments if possible.
• Remember when reading this from top to bottom that the conversation didn't take place in that order. (In a wiki conversation, you can interupt someone retroactively.)
• Occasionally we'll need to stop the conversation, and reorganize it into topics and subtopics so that it's easier to follow.

RfS 08:44, 17 Apr 2006 (PDT)

Some things to consider

We need to look at this from multiple angles:

1. usability for the adopter

   This should be broken into two ideas, usability day to day and administration. Both default Debian and Ubuntu do well in both of these, though the freekbox3 has problems in both areas and Ubuntu seems to have the edge in the administration area. I am sure many other distributions would be fine in this regard as well.

   On the "Gnome vs KDE" question - I don't think the adopters really care as long as they can find their files and their programs and they don't have to call me too often. My experience with Kubuntu is that the version I tried (5.10) did not have all of the functionality of the gnome version in automounting, etc.

I think Gnome is the single most elegant, newbie-friendly (yet capable) interface I've ever seen. It's certainly got Mac and Windows beat for basic use. I would be happy to see a switch to Gnome, if it doesn't cause too many problems with tech support etc. -Pete

I think Gnome is one of the more difficult to use, newbie-hostile interfaces available for linux that's a serious contender for the desktop. this is based on sitting about 5 relatively inexperienced users in front of a recent Gnome desktop and watching their brow furrow and eyes squint. Vagrant 02:59, 19 Apr 2006 (PDT)

1. maintenance of the distribution

   This is really a lot of work, especially when there are major changes in the underlying distro, such as switching from Woody to Sarge, or from Breezy to Dapper. This depends a lot on who is willing to do the work. Vagrant and others have put in a lot of work on Debian. Dave is currently putting in a lot of work on Ubuntu. If we switched to yet another basic distro, we would need to start over on some things.

   most of the work i invested in debian would be applicable to any debian-based system that didn't already implement those features. i've tried to make most of my work somewhat generic. never cling to something that isn't working solely because you've invested energy into it. Vagrant 03:09, 19 Apr 2006 (PDT)

2. tech support

   From my perspective (bows) this is where many of the alternatives get thrown out. Your tech support person in Portland (me) has been using Debian for over 8 years and while there is a lot I do not know, I am totally comfortable adding and removing programs and finding my way around the filesystem. Further, all of the infrastructure machines here run Debian and there is a tremendous depth of knowledge about the distribution. The only reason I am willing to consider Ubuntu at this point is that A) It is based on Debian, so the underlying system is very similar, and B) Dave has expressed willingness to back me up on the parts of the system that are different. Other linux distros have their merits, but if we went that way, tech support would have a tremendous catchup job to do.

3. security

   I am not a security expert, nor do I play one on TV, but the security model in Ubuntu horrifies me. Sudo is a great way of avoiding the inherent problems of letting people log in to root, but the way that Ubuntu uses it, the first user (the one we call guest) essentially IS root. This is very much like the flawed model in MS windows. Perhaps someone with more security knowledge than I have can address this. In terms of security from the adopter doing something stupid with his box, I think it is ok, but in terms of security from outside attack, I think it stinks.

   we would need a process for adding the first user to the "admin" group, and then removing guest from this group. alternately, we could create an "admin" user as the first user, and then tell them that they could add other user's to the admin group to get rootly privledges. Vagrant 03:09, 19 Apr 2006 (PDT)

-- MW 18:31, 17 Apr 2006 (PDT)

Ubuntu security questions

So guest on ubuntu is essentially root like (via sudo). (See Michael's comment above.)

Are the new users who are added also granted sudo access as well (by default)? If not, how do they access root priveledges when they need them?

What happens if the guest account is removed?

RfS 08:35, 18 Apr 2006 (PDT)

I know one of the answers. By default, the subsequent accounts do not have administrative privileges, but they can be given them by the first account, either on creation or later. -- MW 10:25, 18 Apr 2006 (PDT)

I like Ubuntu's security model. Of course, a user who doesn't comprehend it can result in an insecure system...but that is true of ANY security model, so shouldn't be a basis for decision. Ubuntu's model is easy to comprehend, which is a big plus in my book.

In fact, Ubuntu handles exactly how Mac OS X does, so I'd say the lack of a massive outcry about Mac indicates that I'm not alone in this view.

The big problem with Windows security is that you can do all kinds of things from an account that is "administrator-enabled" WITHOUT authenticating at the time of doing it. The user can install programs or delete system files, and so can a script acting as the user...without typing in a password. I know of no other (modern) operating system that shares this wrongheaded approach, certainly not Ubuntu.

-Pete 11:05, 18 Apr 2006 (PDT)

there is a window of opportunity (5 minutes?) where the user has authenticated to sudo and doesn't have to re-authenticate. this is still not nearly as bad as the windows model. overall, i think the sudo stuff is actually a decent way to do rootful things from the end-user perspective. Vagrant 03:13, 19 Apr 2006 (PDT)

Easy To Use Is Poorly Defined

until we come to a well-defined, commonly understood definition of "easy to use", could people desist from calling something "easy to use", and rather describe the qualities and features that they think will meet the needs of the end-user (also poorly defined) Vagrant 02:59, 19 Apr 2006 (PDT)