Should DSL-N start moving forward?
Topic: Should DSL-N start moving forward?
started by: lucky13
Posted by lucky13 on July 28 2007,16:56Safe to say this is UNOFFICIAL. With the discussion over kernels and hardware in the DSL 4 thread, I'm just curious how many people want DSL-N to continue and the direction it should take. And how soon. Feel free to add comments.
Posted by curaga on July 28 2007,17:52I haven't personally used DSL-N, so I don't feel like voting..
My opinion would be that if there's need for it, it should turn to community development. I don't dare to say anything about the direction though..
Posted by Andreas on July 28 2007,18:50Tried DSL-N a while ago but lost interest because of lack of software.
Apt-get would be an argument for me to try it once again!
Posted by lucky13 on July 28 2007,19:18
Just out of curiosity, why apt-get (especially if you mean over MyDSL extensions if they were available and instead of Debian or *buntu)?
Posted by Andreas on July 28 2007,19:21Just because the time I tried DSL-N there have been not so much useful extensions. On my DSL machines I´m using a Debian type (HD-) installation. So there a many more programs to use.
Posted by stupid_idiot on July 29 2007,08:55I voted for:
Say goodbye to DSL-N.
< Puppy Linux > is probably a better GTK2 environment. It is better-supported (It is updated much more frequently, and it has a big user community) and it has tons more software.
Technically, Puppy is the better choice for someone who wants GTK2 and wants it small.
The question is: What can DSL-N offer that Puppy can't?
Posted by lucky13 on July 29 2007,11:36
Security. Puppy runs as root. DSL-N doesn't (not exclusively).
Bloat-free philosophy. Puppy is small on the CD. Take a look at its packages. Puppy's philosophy isn't to remain free of bloat. DSL-N's is.
Want more reasons? The root-only is enough to keep me from using it. I left that behind when I went from DOS to WinNT, which was the first Windows version to separate system administration privileges from normal user privileges. I'm not going to revert to using Linux in a manner I found objectionable, impractical, and insecure in Windows. And that's what Puppy is -- a version of Linux that functions more like Windows 95 than Linux.
Posted by Juanito on July 29 2007,12:58
I don't think it's central to this discussion, but apt-get and many of the DSL extensions work fine for me in DSL-N.
Posted by lucky13 on July 29 2007,13:57
Actually, it should be part of the discussion. Should DSL and DSL-N be as compatible with each other as possible or should they both be totally separate? And the choice about apt-get can be interpreted broadly (encompassing Ubuntu's repositories) or more narrowly to a particular Debian repository (e.g., Woody, Sarge, testing, etc.).
Posted by Key on July 29 2007,14:58Sorry, I hit the wrong button and did therefore a null-vote.
I would have voted for:
"Yes to at least one of the above or to something else, and it should be community-developed if the developers don't want to move it forward themselves".
With DSL 4 alpha and the 2.4.x kernel, I have problems with booting the DSL live-CD from an SATA-CD/DVD drive and also I do not get an automatic internet access with my new computer.
I assume, that my new hardware is not supported from the 2.4.x kernel.
The only thing I want CURRENTLY is a working DSL.
Does not matter if it is DSL or DSL-N.
But it should have a 2.6.x kernel.
The current DSL-N image is now nearly 1 year old.
My mouse does not function with DSL-N 0.1 RC4, but my mouse functions well with DSL 4 alpha, therefore I think there is no problem to fix this issue with a newer DSL-N version.
I know that there is high priority for DSL and the 2.4.x kernel, but I hope that I do not have to wait further months till there is a DSL/DSL-N that is working on newer computers.
Yes, there is Puppy and there is Slack etc etc ..
But also when they have common properties, there are differences.
I had a look to a previous version of Puppy.
I did not like the design of Puppy and some other things.
I hope to get a working DSL/DSL-N again.
I think all I need is a 2.6.x kernel
At the moment, DSL lovers with very new computers can only use a DSL-N version which has not been further developed (at least no new images) since months.
The look of DSL-N should be the same classic look as in DSL.
I would prever an easy blue background with some color structure instead of having a real picture in the background.
Posted by lucky13 on July 29 2007,16:04Thanks for your input, Key.
That's a user issue. You can change the background easily to suit your needs. In DSL-4,
1. Right click and scroll to DFM for X11.
2. Select desktop-options.
3. Select "color" from the dialog near the top (EDIT: left) corner.
4. Set the color using the color tool.
In other versions of DSL, you can use wallpaper in the setup-desktop settings (it has a call for the same color tool if you want a solid color instead of wallpaper). Alternatively and regardless of which version, you can open a terminal and enter the following using whatever color you want:
where colorname (dodgerblue, skyblue2, etc.) is any of these supported by X:
< http://www.imagemagick.org/script/color.php >
Blues are okay. I prefer greys:
< http://lucky13linux.wordpress.com/dsl-related-pages/dsl-40-screenshot/ >
Posted by Key on July 29 2007,16:55Yes, thats clear.
Nevertheless, thanks again for the information.
I was only wondering, when I had a look to DSL-N 0.1 RC4 and found a real picture in the background.
It looks really nice, yes.
But a simpler picture gives with an easy color structure gives a better overview, a better contrast and a better readability.
This should have only been a small remark, as it is in the iso-image and therefore then on the standard live-cd.
On the usb memory stick or the harddisk, it can of course be changed afterwards.
In my opinion, DSL and DSL-N could differ as much as possible and required, but the design should be very similar to either DSL 3.x or DSL 4.x and I do not remember that there was a real picture somewhen in a standard DSL
Posted by stupid_idiot on July 29 2007,17:35
Bloat-free philosophy: Maybe Puppy doesn't explicitly espouse 'bloat-free' as its philosophy, but I think it is also an important part of their objective.
I think that one of the main problems of Puppy is the difficulty of setting up a build environment:
With DSL, I 'debootstrap' Debian Sarge on a partition, and then install the same libraries as DSL with 'apt-get'. With Puppy, I don't think you can do the same unless you have the skills to do a 'Puppy-from-scratch' kind of thing. You probably need to be running Puppy to compile software.
With DSL, I can play games and surf the web on my Debian Sid installation and still compile software in my free time by chroot'ing into the Sarge partition. I only reboot into DSL when I want to test extensions.
I think Puppy's dilemna is manpower - they need more people who can be familiar with the system - libraries and so on - and have hardware that's fast enough to compile GTK2 software full-time. Yet the complexity of GTK2, in general, drives people away. So the 'task' has gotten more complicated, but there are less people who are willing to do it. But, despite this, Mr. Kauler seems to me a very motivated developer. If he keeps improving Puppy's 'core' as time goes on, then I think more people will be interested in maintaining the packages. Then the level of quality will go up.
You said that Puppy reminds you more of Windows 95 than a proper Linux distribution - in this regard I totally agree. What I think is that Puppy resembles '95 not so much in the security problems as in the desktop experience - this I think is a positive boon rather than a harmful thing. Let's assume for argument's sake that it is very unlikely to get Linux infected with spyware and viruses - the 2 big killers - whether you run as root or not. The remaining security risk is running as root itself - which I think deserves a debate in itself.
I have read this before: < Slashdot | Sudo vs. Root >.
There must be many other discussions about this controversial issue.
DSL-N will end up competing directly with Puppy on availability of software, and then, consequently, ease of use.
It is not that the ideas for DSL-N are bad. But in the end, I think both works have to serve their niche, and do it well.
It is that competition in these areas will be divisive. I feel DSL needs people and time on the 'small and simple = better' side, and Puppy needs people and time on the 'more and latest software = better' side.
Posted by Key on July 29 2007,18:21" You said that Puppy reminds you more of Windows 95 than a proper Linux distribution - in this regard I totally agree ".
>> And this is one thing in Puppy I do not like.
Also computer development continues.
It would be a pity, if there are more and more computers in the future, where DSL can not be used on due to not supported hardware in result of an old Linux kernel.
DSL should expand its size limit maybe in reference to the available technology.
All the time (how many years?), it has been set to 50 MB.
Why not do 80 MB or something like this now?
I am sure Puppy can be underbidden with a 2.6.x kernel as well.
Probably the kernel 2.4.x has some other advantages to the kernel 2.6.x, than only its size and therefore it should and could be continued as well.
Nevertheless there should also be an up-to-date DSL version for the latest hardware.
I do not want to switch to Puppy
Posted by lucky13 on July 29 2007,19:08
I don't. Users who want a similar desktop experience can use JWM and dfm like DSL 4 is using (dfm is much more like Windows than rox). I don't consider running as root a viable or reasonable compromise, and I can't advocate people who want a 2.6-based live CD to install to hard drive choose something like Puppy. People who presume that Linux is inherently safer than Windows are greatly mistaken, especially when they start from the same user=administrator baseline Windows 95 did. Windows 95 wasn't inherently unsafe because it had a registry, it was inherently unsafe because it was wide open by default.
SO IS PUPPY. SO IS DYNE. SO IS ANY OTHER DISTRO THAT ABUSES/MISUSES PERMISSIONS AND RUNS EXCLUSIVELY AS ROOT TO MAKE THINGS "EASIER" FOR USERS. IT ALSO MAKES THINGS EASIER FOR EXPLOITS TO THE WHOLE SYSTEM REMOTELY -- AS WAS THE CASE WITH WINDOWS pre-NT -- AND LOCALLY. THIS SHOULD NOT BE AN OPTION, AND THIS IS WHY PUPPY SHOULDN'T EVEN BE CONSIDERED "LINUX" OR CONSIDERED IN ANY DISCUSSION ABOUT THE FUTURE OF DSL-N. DSL-N IS SAFER, IS A BETTER IMPLEMENTATION OF LINUX.
(Sorry for caps, but puppy:dsl-n is totally apples:oranges.)
I'm not going to make that assumption, and I'm also not going to assume that running as root is as inherently safe as using sudo. It isn't. Running LOCALLY, on a non-networked computer, as root is relatively safe if you know what you're doing. Running on a network (internet) as root is NOT safe. You can dismiss it on the grounds that Puppy (or Dynebolic or any other CD that runs only as root and is installable) is read-only, but anyone able to exercise control of your system locally or remotely can mount any partition and do whatever he or she wants.
And with the proliferation of live CDs that run as root (puppy, dyne), it wouldn't be difficult for a server that's either run by unsavory people or has been compromised to take advantage of people using them (deleted fuller concept). Your worst nightmare then isn't a virus, it's someone getting a dump of your hard drive's contents, erasing it or your MBR, and/or even doing something that can potentially harm your hardware.
Another big difference between root and sudo: you can't tighten permissions on root. You can tighten them for other sudoers. I think you SHOULD. Using sudo shouldn't be indiscriminate or capricious, it should be thoughtful and methodical. I think it's wise to limit what applications -- particularly those that reach networks -- can do when used as a user with sudo privileges. For example:
< http://www.xs4all.nl/~hanb/documents/firefox_for_paranoid_people >
I wrote within the last couple weeks in another thread that I usually scrap sudo (if it's part of a default install) or don't install it at all (e.g., on a BSD install). I'm not a fan of it even though I agree it's a reasonable compromise, particularly where there's a need for more than one person to be able to make system-wide changes. It prevents the need to give root passwords to more than one person. It's also auditable. Etc.
I think you're too dismissive of the risks and too permissive of puppy's sloth in the same regard.
EDIT: I don't mean for this to sound so harsh and it's not personal, but I think the issue is too important to treat lightly or to treat as "they do it this way" and "we do it this way." They do it wrong. It may not be a major problem right now, but the potential is there for widespread harm. I don't think developers should accept that as a trade-off. I don't think users should accept it, period.
Posted by lucky13 on July 29 2007,22:57Rather than have this post turn into a debate about merits of different distros, let's refocus on the issue at hand and get some input on how this should go forward if it should at all. Assume that it will have 2.6 and more support for newer hardware, assume that it will receive the kind of development DSL itself has. What do YOU want out of it? If the choices above don't suit your concerns, please provide feedback.
Posted by roberts on July 29 2007,23:15Lets try to stay on topic and positive here. While I may agree, I don't want to start any distro wars.
To the subject at hand:
I am not sure how to respond to this. Am I to take this as an affront to the result and conculsion of the very recent poll which resulted in a consenus of how DSL was to move forward? And now, when work has only just begun on a new interface and improved user experience?
I have many times posted why dsl-n development has stalled. The main of which was lack of community interest. Where DSL forums had an average of 100 users, DSL-N had 5. Five. Where do you think I should have spent my time? Follow that up, with the recent poll and there was not an overwhelming demand for a larger 2.6k DSL. DSL has built its reputation on supporting very low end hardware, the Libettro, and 32MB-64MB machines.
I realize that over time things change, but I am talking very recent. dsl4 is only at alpha1.
I also realize that some in the community want new hardware soultions ASAP.
Given all the above, and if I am not being excluded from this discussion. This is what I would propose...
1. Evolutionary not revolutionary, i.e, do not try to maintain two separate distributions.
2. Leverage what I am currently developing.
3. Leverage existing gtk1 extensions.
4. Leverage existing gtk2 and gtk2 extensions.
I am swamped with work on dsl4, but if one or several in the community, wish to compile a minimal 2.6 kernel, modules, and the core third party modules needed by DSL( ltmodem, unionfs, cloop, ndiswrapper, fuse, prism2, and madwifi). I will incorporate them into another "edition" of DSL.
Most likely this kernel and boot time modules, ide, cloop, usb, and sata will not fit on a single floppy disk. Therefore likely no syslinux edition. Most likely only an isolinux edition.
I would work to automate the process of simultaneously releasing both a 2.4k and 2.6k editions of isolinux DSL.
By adopting this approach, we would leverage current developemnt, use existing gtk1 extensions, easily add gtk2 and gtk2 extensions.
We would offer greater new hardware support while minimizing operational support questions and issues.
This would at least be a start on which we could later begin to update and/or replace core applications and libraries.
Any volunteer kernel builders?
Posted by globalsize on July 30 2007,04:39Well, I have a dangerous habit of getting involved in projects that are time-consuming and beyond my knowledge....
But hey, if someone is willing to help me out, I wouldn't mind trying to help DSL-n get back on its feet.
If this is going to be another edition of DSL, would that make a 2.6k version something like "DSL-n 4.01RC1"? After all, it will be the same design as DSL4, just, y'know... different.
@Roberts: You have my full appreciation, so don't think that the re-interest in DSL-n is a violent populist upswell. With the Little OS That Could, my Satellite is now enjoying a robust second life. The reason that I personally liked DSL-n is the addition of gAIM (now Pidgin), as well as the fact that I didn't have to change my BIOS settings to have DSL not hang when it searched for my PCMCIA card on boot.
Posted by lucky13 on July 30 2007,04:51
Not at all. I'm doing this because:
1. I know you're very busy with DSL4.
2. The issue was raised about new hardware not being supported much further in 2.4 and I thought the dialog needed to be moved away to a more appropriate thread.
3. I want to see if there's a consensus among those who do want DSL-N developed about what they direction they want it to go.
4. I want to see how motivated they are to get it rolling.
You're definitely not excluded.
Per your four points, I included in the poll specific choices dealing with parallel 2.4 and 2.6 development. I laid aside my own presumptions that those wanting 2.6 would also prefer GTK2 when the issue was raised.
I think isolinux-only is sensible for 2.6 since machines requiring syslinux will likely find their hardware better supported in 2.4. Am I off on assuming that?
Posted by roberts on July 30 2007,18:01
Correct. However, during the DSL-N forum days, there were many who squawked about this fact. My feeling is that most newer hardware does not even have a floppy drive. And if both "editions" , 2.4k and 2.6k offer the same operational features then not offering a syslinux/boot-floppy for 2.6k should not be a problem.
Posted by Key on July 30 2007,18:53
I think the same, Robert.
If there wouldn't be a problem with the hardware support on newer computers, I would be happy with the 2.4.x kernel. But as the hardware support is limited, I'm probably forced to use a 2.6.x kernel. I hope everything is working then again with kernel 2.6.x on the new computer.
In order to reduce the time and work, I would also be happy with an ordinary DSL having the 2.6.x kernel.
People with older and newer computers are able then to test DSL4. Also I assume it would be the first distro supporting both kernels for the same version.
An ordinary ISO image would be fine.
Boot floppy and special ISO images are not really necessary for kernel 2.6.x users.
Posted by lucky13 on July 31 2007,00:24globalsize:
That's available via apt-get as well as a GTK2 version in MyDSL. You don't need DSL-N for gaim.
I agree. It shouldn't be the same kind of issue if there are two kernel versions of the otherwise identical distro since DSL-N was totally different with GTK2, different apps, etc. It's just pick the CD that best supports the kind of hardware you have. And give a really big hint that if you can't boot from CD, you don't need 2.6.
Posted by humpty on Aug. 12 2007,04:07I would like to see the only diffference between DSL and DSl-N is
they have different kernels. And nothing more.
Even better would would be if DSL-N could be replaced by LKMs (loadable kernel modules). Though, you must realise I'm just talking
about stuff I know little about now, but I think linux made a design
hiccup not making the majority of drivers modular.
Posted by lucky13 on Aug. 12 2007,05:41Linux already uses LKMs. There are three choices for handling drivers when compiling a kernel: drivers can be compiled directly into the kernel, compiled as modules (which are then turned on or off as needed), or not compiled at all.
The first option is ideal if you're building a kernel for use on one particular machine. You can configure it so it's machine-specific and doesn't have any modules at all (or only a few in the event you have hardware that you may add or remove on occasion and if you want those drivers to be loaded/unloaded as needed). That would also pertain to the last option -- you would leave out drivers you don't need for that machine. The result is a very streamlined kernel.
Modules are a middle ground, and that's how DSL handles drivers. Since DSL is to be used by many different users on many different machines, it includes drivers as modules that are turned off and on by users or configuration scripts. You probably don't even scratch the surface of what modules are available because DSL's kernel is compiled with more modules than most other distros. You boot up, the modules you need should load (or you can insmod); the modules you don't need don't load (or you can rmmod if they do). This also results in a rather streamlined kernel insofar as your measure is what's loaded -- it should be what's germane to your computer. The flip side of it is that you have a lot of modules available that you won't use -- dead weight to some people, but that's the price of using something that's designed for wider appeal than one particular computer (it's a live CD!).
The only other choice DSL has would be to compile more directly into the kernel and have fewer modules. That would result in one massively bloated kernel with excessive "dead weight" for everyone because we all have unique needs.
Posted by humpty on Aug. 13 2007,08:06Is it possible to offer the LKMs as an extension and to load them only afterwards?
That way the base distro could be light weight.
Posted by lucky13 on Aug. 13 2007,12:22
The base distro is already light weight. I don't think moving modules to extensions is the best way to make it lighter. Just look at the aggravation people who need the ipw2* driver extensions face. I don't think people want to have to download one driver for one old soundblaster card on one machine and then another driver for another card on one machine just to save a couple MB on the CD.
I don't see how (or why) you'd want to do that en masse per what I wrote above about how it's a live CD intended for a wide audience. Looking at it from a cost:benefit angle, I think it would be too much of a hassle to start removing the modules from the CD and make them to extensions. The CD wouldn't be useful as a CD until people downloaded the drivers they'd need.
You could do that on your own remaster with your own kernel if you wanted, or you could do a remaster suitable for more specific hardware configurations. But for a general audience, it needs to have a wide variety of modules so it can be as one-size-fits-all as possible.
Posted by lucky13 on Aug. 13 2007,18:03Piggybacking on earlier thoughts...
The more I think about it, the only place it makes sense to pare down the available modules is in a machine-specific environment either narrowly for your own computer or more broadly for a particular model of computer like a certain brand of thin client where additional modules will never be needed. This is already something accomplished best via remaster, and in those cases it's probably better to compile the drivers into the kernel anyway (no modules) because it's for limited use.
Where it doesn't make any sense to me is for those who will use the live CD or a USB stick promiscuously on any number of computers -- and these are main uses of DSL. Moving the modules to extensions to lighten things by a couple net MB would require people to know exactly which modules they'll need ahead of time so even the CD can be run properly.
It also wouldn't work well for those of us who shuttle hard drives between very different computers. As it stands now, I can move any of my hard drives around and run xsetup and DSL will work on any iX86 PC I have here. I'd have to keep multiple configurations to either move my hard drives around or to use DSL-on-USB on different computers so the different extensions would load when I need them.
But my main thought is about the people who will download the CD just to give it a spin. Some of them have a difficult enough time as it is adjusting to DSL from whatever OS they were using before. I imagine it would be a deal-breaker if they had to figure out every single driver they need and then load them manually just to get things up and running.
Posted by curaga on Aug. 13 2007,18:38Sorry, but you seem to be repeating yourself for 3-4 posts..
Posted by lucky13 on Aug. 13 2007,18:39One's repetition is another's elaboration. :-P
Posted by tinker on Aug. 14 2007,13:32It's difficult to get significant results in a poll, especially in an unofficial poll. That difficult is increased if one who runs the poll continues to let their own preferences show or who discounts the explanations that were asked for. You would probably get a better idea of what people want, or what they think they want, (and more people would be comfortable discussing it) if you let this go as a "brainstorming" session and then discuss the results and whether or not the suggestions would work after the poll closes. However, it's your poll so you can run it as you choose, I just wanted to mention this for your consideration lucky13.
Posted by lucky13 on Aug. 22 2007,17:24
I appreciate your concern, even though I wonder where I ever expressed my opinion about the issue related to the poll. The opinions I offered were about other issues: comparisons between DSL and another distro, handling modules as extensions versus in the base distro, etc. I also think the poll became a moot point when Robert floated the idea of making DSL 4 available with two kernel versions, and then saying DSL 5 will be 2.6-based.
Regardless, it's clear from the fact there are fourteen votes that DSL-N lacks interest from the rest of the community.
Posted by K'net on Oct. 15 2007,06:29I know the conversation is past this point, but I would just like to voice my own opinion on the matter of continuing DSL-N in general.
I believe that DSL-N is a very worthwhile project. I started learning about Linux a couple years ago and DSL-N has helped show me what sort of things are at the core of a linux distribution and basic methods of manually installing some applications. Being a college student I don't have a lot of money to go out and buy the latest hardware, so I appreciate something I can learn from, while at the same time being usable.
I hope that you will continue this thread in a positive direction that will eventually lead to a new edition of DSL-N. I only wish that I knew enough that I could help.
Posted by Key on Oct. 15 2007,16:57
If you can't afford the latest hardware, then you should be quite happy with the ordinary DSL 3.x and 4.x with a 2.4.x kernel.
What is the reason why you want to have a new DSL-N with a 2.6.x kernel ?
Is your current hardware nor supported in the 2.4.x kernel ?
I had these problems and have switched to TinyFlux / TinyMe in the meantime. Which is running quite well on my new computer.
Nevertheless, I still like DSL and was always waiting for a version that is working on my new hardware.
But I respect that all the developing capacities are set to the DSL 4 with the 2.4.x kernel and I understand that this is a lot of work for a one-man-crew.
Let's wait till DSL 4 is ready and maybe then Robert will continue releasing a version with a 2.6.x kernel (DSL-N or DSL 5 ).
Posted by RigasW on Oct. 15 2007,18:07Hi Everybody,
I used DSL for about two years. Two month ago I found Zonbu. Zonbu is a Gentoo based Linux that runs on a 4GB Compact flash card and stores the user's data on a server in the internet.
OS and apps are modern, up to date and well maintained. Os and apps are patched automatically about once a month.
Have a look:
< Zonbu Homepage >
Posted by roberts on Oct. 16 2007,04:07The new DSL-N or DSL v5.0 will come after 4.0. I have started to assembly some ideas based on user requests. It will likely have a different base but still remain Debian compatible. But I am only one person and cannot juggle so many versions and editions of DSL. Also I don't want to be another one of so many similar 2.6 based small distros. I am thinking very minimal. But hey, that's not new. Cannot commit to any date. Just wanted to let eveyone know that it is being more than contemplated.
Posted by Juanito on Oct. 16 2007,06:24Good news for users of newer laptops