Current Issues &Forum: Site News Topic: Current Issues & started by: roberts Posted by roberts on Jan. 15 2006,17:25
Current Issues & Road Map AheadJohn and I have been discussing the current issues that have appeared in the forums. We wish to share with the community our plans to deal with them and what's ahead. We knew by moving to 2.4.31 kernel that there would be many challenges, especially since it is to difficult to know in advance that all existing hardware would be well supported. Having very limited resources to test such, we could only build it and let the community give us feedback. Which you have provided. It is apparent from feedback in many of the forum threads that we have lost more hardware support than we would have expected. It is also apparent that we have not gained much at all with this move. We thought that by providing full sources the community would help in building the necessary needed modules to support missing or lost hardware devices. This too has become apparently not such an easy task. The final incident occurred when we learned that the atmelwanlan drivers requires a non smp kernel. This is something that is totally outside of our control. It is our conclusion that this move to 2.4.31 kernel is trying to accomplish too much. By trying to maintain support for older hardware and at the same time trying to support some of the newest hardware, we end up with a hybrid. One that apparently does not do a go job of supporting either old or new. Therefore, we have decided, that the next version of DSL v2.2 will be based on the kernel and modules of DSL v1.5. Most of the feedback has been "that it worked in 1.5". To address the issues of the newer hardware, we are announcing an additional version of DSL to be based on 2.6 kernel. The exact date and exact size of which is to be determined. We also have in planning an even smaller version of DSL which will be based on 2.4.26 kernel. We appreciate the community in providing feedback and helping in the many ways you do. John and I have much work ahead to provide a better DSL experience. We both look forward to your continued support and participation. Posted by mikshaw on Jan. 15 2006,20:03
So I assume the old nvidia driver will once again work in 2.2?What about ltmodem support? If not, I think i still have Stinky's mydsl package he made for DSL1.5 The "even smaller version of DSL" sounds like something I will really enjoy. Posted by newOldUser on Jan. 15 2006,20:23
Sounds like a good idea. It might become tough for you to juggle updates to multiple versions as time goes on but this sounds like a resonable starting point. Perhaps there is only need for two or three 'dot-something' updates for each version per year. No matter what... Thanks go out to you, John and the entire DSL community for a fun ride so far. Posted by starcannon on Jan. 15 2006,22:12
You guys are awesome thanks for a great distro to use on my old 366mhz trashtop. It had dust on it before DSL came along, now its my favorite puter for email, internet and music... i have a patch cable that lets me jack it into my nice home theatre and let xmms rock the house Anyway hope maintaning 2 seperat forks is not to overwhelming, I'm noob, but I'll keep workin on my skills and see if i can't contribute something someday soon. Rob Posted by MethodOne on Jan. 15 2006,23:31
Will there be any changes in the apps of the versions for older and newer hardware?
Posted by roberts on Jan. 15 2006,23:56
we are planning that the three versions mentioned will each have a different collection of base apps. Still keeping size in mind. We have no size limit currently planned for the 2.6 kernel version other than ASAP (as small as practical). We plan on maintaining and expanding our unque DSL architecture. That being compressed type installs and easy installation of additional extensions. It is an ambitious plan but one we feel is best instead of trying to stretch 50MB to fit both ends of the spectrum.
Posted by mikshaw on Jan. 16 2006,00:23
Here's another possible issue....mydsl packages. If the different DSL types have different kernels, some packages will work in one and not in another.I assume *most* packages will work in all DSLs, as long as the libs are the same (or at least compatable). Just wanted to let you know that on dialup I will probably stay updated with the smallest version only....any mydsl packages i put together will probably be built for that version only. Posted by roberts on Jan. 16 2006,00:32
We are not planning on abandoning the smaller version(s).It will most likely remain those that will need special packaging of custom extensions. I will look forward to your continued participation. My first order of business is to get v2.2RC1 ready for public testing. Posted by doobit on Jan. 16 2006,02:10
2.1 is working extremely well on my 233MMX laptop, but so did 1.5. Either way, I'll keep learning DSL until I become proficient on it. I hope that way I'll be able to contribute more to the community.
Posted by dare2dreamer on Jan. 16 2006,06:10
As applications and versions are starting to be tossed around, is there any chance netrik could be removed and links (or some decent variant) put back in?While netrik may be smaller, from a functionality standpoint it's miled behind DSL's original text browser options. Posted by cbagger01 on Jan. 16 2006,07:10
I think that there are more "happy customers" with kernel 2.4.31 than it seems.This is because you usually only hear feedback when something goes wrong instead of when something goes right. This non-smp thing stumps me, because DSL 1.5 comes with an SMP enabled kernel. I see two penguins when booting on my HT Pentium 4 machine. But if the need to go backwards is still considered to be a must, I would like to make the following request: Instead of jumping back to 2.4.26 / KNOPPIX 3.4 / DSL 1.5, can you instead jump back to a 2.4.27 / KNOPPIX 3.7 base system? This will include the knoppix autodetection scripts for the last 2.4.x version that was supported before the jump to kernel 2.6.x in knoppix 3.8 It will also include SATA support, which will be a big requirement in order for the "Small" DSL to be able to work with just about any modern PC shipped in the last couple of years. So the end result might be: Some newer hardware is supported SATA is supported More up-to-date KNOPPIX autodetection scripts Still the same stability as the knoppix 3.4/2.4.26 systems Is it worth a try? Posted by pr0f3550r on Jan. 16 2006,10:56
What about a "dsl-devel" mailing list, ideally hosted on lists.debian.org?
Posted by clacker on Jan. 16 2006,13:59
I was very happy with dsl 2.1 and the 2.4.31 kernel. I loved being able to view my SATA ntfs hard drive again (thanks), and the additions and changes Robert and John have made have been great. I understand that drivers needed to be written, but look how long it took to get all of the drivers for the earlier kernel.I think getting two versions working is going to be even more work than getting the 2.4.31 kernel up to speed. It might be best to stick to your target audience, drop the kernel back, and leave it at that. Posted by skaos on Jan. 16 2006,14:16
Is it possible to add a full X-server in the 2.6 kernel version?
Posted by Grim on Jan. 17 2006,02:08
As far as the myDSL extensions are concerned, when they're packaged, why not just add the release number.That is, if a .dsl of mplayer is packaged for 2.1, write the filename as mplayer-2.1.dsl. That way, at a glance, the user will know for which release the program was packaged. That will result in multiple versions of packages if they're incompatible between releases, but if the old package works with the new release, a symlink can point to the old package, or a copy with the new version number can be uploaded. That would effectivly eliminate the incompatible package problems. If I get a vote for text-mode browsers, I vote that links-hacked goes back in. I used links-hacked more than I use firefox (I'm using it right now) and even though it's no longer actively developed, it's extensible with lua, which DSL heavily employs, and even though it's a few megs, it's useful. I'm sure it's not coming back, but I use the hell out of it, and it'll always find a place in my toolbox until something better comes along. The only other thing I was wondering about is whether or not we've thought about using upx or another form of executable compression. Upx will shrink links-hacked to half it's size. Run that on half a dozen stand-alone executables and we might be able to shoehorn another app or two that previously wouldn't make the cut. (I wonder how small nvi and zile would be after packing?) Posted by clacker on Jan. 17 2006,03:15
I created an nvidia.dsl to work on dsl 2.1 and I have submitted it to extensions for testing and approval. I know now it's a day late and a dollar short, but... I fear it's too late to save the 2.4.31 kernel, but I feel I need to try. Posted by clivesay on Jan. 17 2006,03:49
During my recent travels I was told I needed to eat at a hamburger joint called '5 Guys'. I had never heard of it but was told it was THE place to get a hamburger and fries. I walked into a simple red and white restaurant with simple seating. The menu was also simple, basic hamburger, cheeseburger and bacon cheeseburger...all with free toppings of your choice. That was pretty much the extent of the menu. As I sat down to eat, I looked around the restuarant and noticed magazine and newspaper articles posted everywhere praising this establishment for their food. They had taken a simple hamburger and focused on perfecting it. There were no buffalo, turkey, veggie or soy burgers, just hamburgers. When people go there they know what they'll get and expect the quality to be high. Hamburger - DSL isolinux/syslinux Cheeseburger - DSL USB Bacon Cheeseburger - DSL VMX Fries - Lua/FLTK Toppings - MyDSL My Hamburger (base DSL) & fries (Lua/FLTK) are are tasty as is and I can add any toppings (MyDSL) of my choice. I'm afraid Buffalo (2.6 DSL) and Turkey (Minimal DSL) might cause the cooks to be so pre-occupied with other things that my beloved hamburger (base DSL) will suffer and not taste as good. All of that additional expense and time cooking the other burgers may mean that I start to get less fries (Lua/FLTK) with my beloved hamburger (base DSL). That's hard for me to stomach because I come to this place for my hamburger cooked just the way I've always liked it. I don't mind the cooks experimenting with new spices from time to time but I want the basic burger to stay the same. If I have occasional cravings for Buffalo (2.6 kernels) and Turkey (Minimal linux) then I can go somewhere that has been specializing in those to satisfy my taste but I want to be able to always come back to my hamburger, cooked to the highest standards. That's what my favorite restaurant was built on. If I have one critique of my fav burger joint, it's that it decided at some point to remove the person in charge of the toppings (MyDSL) so they are kinda scattered all over the counter (Forum). They used to be neatly organized and if any toppings were getting old or had bad spots, that person would freshen them up and/or remove the bad spots. Right now I kinda have to pick through them and have to search for the toppings I want, spread all over the counter (Forum). Now, maybe my fav joint will hire more cooks to make the other burgers, I just prefer the head cook (RS) keep his focus on that great burger I've come to love. That's my take, now I need to get something to eat Posted by rattler414 on Jan. 17 2006,04:00
Mik, I could not agree with you more. Flavors and variety are nice but the tried and true staples should remain just that..staples. If I want 2.6.x I can use what ever I want ) If I want Koffice or what ever I'll go to a diner that does them well. When I want a solid light meal I'll come back to my "burger joint"
Posted by mikshaw on Jan. 17 2006,04:21
You buffoon! that was chris =o)
Posted by cbagger01 on Jan. 17 2006,06:05
In my opinion, what makes DSL great is the combination of:great hardware detection (slimmed down knoppix core and knoppix autodetection scripts) small and light additions (light Xserver,Window Manager, apps, printing, OPTIONAL light services, etc) DSLness that makes dsl DSL and not just a stripped down knoppix (every concept, script, extension system, user contribution, etc) I believe that the bottom 2 items are and always will be part of the core that makes up DSL. However, I also think that the knoppix autodetection part is very important. The knoppix team does an unparalleled job of evolving (and usually autodetected) hardware support for a tremendous variety of hardware yet while running in a livecd environment. Instead of doing some kinda "Linux From Scratch" project, the DSL developers wisely chose to build on this KNOPPIX foundation, and this is one of the main things that separates DSL from similar projects not based on KNOPPIX. This way most (not all) supported hardware "just works", while other projects need to go through the same learning curve that the knoppix guys once did, and usually with a much smaller team than the knoppix community. Of course, after KNOPPIX 3.7, they eventually dropped the 2.4.x kernel series for the 2.6.x series. So most new knoppix hardware support development would only be useful for a "DSL with 2.6.x kernel" project. But for the current DSL (2.4.x) project, there are tradeoffs to consider for future direction: 1) Go with recent 2.4 kernel build like 2.4.31 + Latest 2.4 hardware support and some bugfixes + Mostly still works with the old KNOPPIX autodetection scripts, but some new drivers may need manual loading + ACPI now works on some hardware that broke it in the past - ACPI sometimes interferes with some hardware/BIOS + SATA SUPPORT IMHO this is a REQUIREMENT for a "portable" OS because SATA is taking over the computer world - Some wireless hardware is now inoperable. An investigation is ongoing with hope for a fix/workaround of some sort but at this point is still just hope. 2) Go with a remaster of the last KNOPPIX to support the 2.4.x kernel, 2.4.27 and KNOPPIX 3.7 + famous KNOPPIX tested hardware support. If this kernel version supports your hardware, it will "just work" + SATA SUPPORT IMHO this is a REQUIREMENT for a "portable" OS because SATA is taking over the computer world - Not the latest kernel, so some bugfixes and new hardware will not be supported. - Unknown if some hardware support was broken when linux moved from 2.4.26 to 2.4.27 - More work. Developers must dig out the old "how to strip down KNOPPIX to make the DSL core" notebook. 3) Revert back to existing DSL1.5/KNOPPIX3.4/2.4.26 system + If your hardware worked in DSL1.5, it will work. + Relatively easy to backport to DSL 1.5 vs. other options - NO SATA SUPPORT This is the biggest hangup for me. IMHO SATA support is a REQUIREMENT for a "portable" OS because SATA is taking over the computer world - Not the latest kernel, so some bugfixes will not be included. Anyone whose hardware "now works great since DSL 2.0" , will lose this functionality. So there are pluses and minuses for each direction. Personally, I have a strong preference for (1) or (2) because being able to see the hard drives on just about any computer built after 2003 is a BIG plus to me because I use DSL as a portable USBHDD system and I don't get to choose the computer model at my friends'/relatives'/work/etc so it is inevitable that I will encounter SATA drives. Posted by RoGuE_StreaK on Jan. 17 2006,06:20
Anyone else hungry?
Posted by dare2dreamer on Jan. 17 2006,06:49
I nominate the "burger analogy" as the new codename system for future DSL releases. If Shuttleworth can have badgers, hedgehogs and a dapper drake, why can't we look forward to the next tasty release of Friendly Frenchfry or Patty Melt? Posted by roberts on Jan. 17 2006,08:45
With the current 2.4.31 based system.we lost scsi boot device support we lost zip drive support. we have issuses with pcmcia support we have issues with wifi support we have issues with usb boot support we have gained sata drive support. Well many may not notice these things, others do. We have lost some users, or at minimum made it very difficult for them to remain among us. This also seems to be a temporary solution. There are other features that I would like to try to take advantage of that only 2.6.x will provide. As for the suggestion of 2.4.27 Knoppix 3.7 for sata drive support. Well it is not a panacea. Many of you do not know of, or wouldn't even be aware of my efforts, until we started the release candidate procedure. But back in May 2005. I created the first 2.0 version of DSL. It was based on such kernel of Knoppix 3.7. I worked on it from 05.13 thru 05.30 with thirteen pre-releases to the then beta testing team. Because of several similar issues to the current 2.x, it was dropped and never released. Again, many issues with the basics of extended sound (ALSA) and others prevented this from being released. Trying to use a single 2.4.x kernel, seems not possible without losing at both ends of the hardware spectrum. After my second attempt to try to deliver such, one private to the testing team last May, and now this publc one, I feel quite beat up. It is not anything under my direct control. It is just the nature of the beast. I know no matter what I do, that I cannot possibly please eveyone. But I also know that you cannot squeeze "blood out of a turnip". That is why the decision was made to use the proven 2.4.26 kernel and modules of DSL v1.5. Everything that is in 2.1 that is possible, including ltmodem autodetection, all the latest lua GUI scripts, will be there and all the original myDSL extensions will work as before. Just as one should never stop learning, so too should one always try to plan for what's ahead. We will plan to work on a 2.6 kernel version for the newer hardware based on Knoppix but with a very similar feel to our beloved DSL. We will continue to improve DSL. That is the best that I can offer. Posted by frankseu on Jan. 17 2006,11:51
Hello Roberts,i can only immagine the problems you and the other devs have. I am very happy with DSL an love the work you do. I am also happy to read that you and the other devs still have open eyes and a open mind for changes/wishes. I often see a kind of blindness against the own project of the devs after a while. You and yor colleagues are not moving into that ! For that i like to thank you and wish you power, time, endurance. (And a tolerant wife ;-) regards Frank Posted by rattler414 on Jan. 17 2006,12:31
Mik, don't pick on my use of your name instead of Chris's - I had some Baileys and was feeling very good
Posted by clivesay on Jan. 17 2006,13:50
CB -I completely understand the issues, especially SATA. As Robert said, 2.4.27 had it's own problems too. It seems 2.4.26 is just the most stable and far as the best support. Actually, though, the reason 2.4.26 is comfy for the community is because of myDSL. Extensions are broken for other versions. This is not a dig at the dev, Robert has all he can do. The fact is, though, is it needs to be decided if MyDSL is going to be officially supported or not. If not, then let RS develop the base and alsa and everything else may or may not be there, that's just going to be the way it is. If MyDSL is officially supported there needs to be someone to manage it. The newer builds of DSL are requiring things like updated nvidia, alsa and gtk2 libs. To do that type of work takes skill and time to do it correctly. DSL has gotten away from that and now some of the 'staple' extensions no longer work and extensions are not being submitted. There are 'unofficial' ones popping up everywhere. So for the future of DSL are things like sata and wireless the priority? If so, let that be the focus. If sata, wireless and mydsl is to be supported, then some decisions need to be made. If Robert decides to do a 2.6 version to support newer hardware, that's great for the hardware but adds yet another level of complication and management to MyDSL. Robert mentioned that some users seem to be moving on. I would guess the fact that extensions don't just 'work' anymore maybe part of the reason. This is my own perceptions. I personally use a 1.5 base because my hardware is well supported, including nvidia and such and if I want to compile an app I can still use Knoppix 3.4 without alot of grief. I need that crutch because that's the extent of my linux knowledge. Posted by roberts on Jan. 17 2006,16:10
I take issue with CL's claims regarding extensions.The myDSL system is a community process. As such it is not nor should be the responsibity of the devs of DSL. I created the myDSL infastructure to allow this third party software packaging to allow users, the community, to participate in the process. That is similiar to third party software that one would obtain for any OS. We have setup a separate web hosted area (the repository) for them and it clearly states that they (the extensions) are put together by users of the community are separate from the DSL project. We have setup a separate section of the forums so extension makers and users could have a dialog on such. Whatever, changes I or John, may make in the base, whether it is kernel/module related or not, may indeed require an adjustment in some if any extensions. Is that the responsibilty of the devs of DSL. I think not. Would you go to the OS maker for support of your third party software? To make such a claim, would be like saying Microsoft cannot go to Vista, because not all third party software would work. I find it unfornate that the purpose and structure of the extensions would come up and need to be re-explained. The extensions are not the reason. Supported hardware spectrum is. Posted by guest on Jan. 17 2006,17:50
Since when? I seem to remember that there was one (ke4nt1?), or maybe two people who managed the repository, and spent quite a lot of time rebuilding packages when things in DSL broke them, and managed new submissions. MyDSL has never been a community led/run/supported/whatever process. Yes, users submitted extensions, but I figure they were usually of such poor build quality that they had to be completely rebuilt to function properly. Posted by cbagger01 on Jan. 17 2006,18:08
Robert, I was unaware of the DSL/knoppix 3.7 RC testing period, so I can't comment on the issues encountered other than the ALSA one. I did notice that when booting a stock KNOPPIX 3.7 livecd, you need to specify the ALSA driver name in the boot cheatcode ( knoppix alsa=snd_mydriver) because the default boottime autodetection was suspended for some reason. As for the other DSL 2.x issues, SCSI boot support is removed for space issues and can be addressed by the user with the modules directory zip drive boot support can be addressed with a custom boot disk like was done in older versions of DSL I have a hunch that USB boot support can be improved by tweaking the linuxrc and maybe add a few cheatcode options. For example, maybe a cheatcode "nousb2" to disable the autoloading of ehci-hcd can be added. But the big problems are in the area of pcmcia and wifi support, for which there is no clear solution in sight. However, I would hope that jumping from 2.4.26 to 2.4.27 would not break these like the jump from 2.4.26 to 2.4.31 did, especially since knoppix went through a few rounds of bugfixes in the 3.7 series, just like the 3.4 series once did. As for beating up the livecd maintainers, these people need to get a life. Other than the removal of SCSI modules to stay under 50MB, no intentional effort was made to "break" hardware support for anyone. The upgrade was done using a nearly stock linux kernel, so no kernel hacking tricks were done to break support for hardware. In fact, the "compile your own driver" process was made easier by the packaging of the appropriate sources, configs, patches, and compiler tools vs. the old days of "Download the entire KNOPPIX 3.4 5-17-2004 livecd" and go to work. It is inevitable that some things won't work after making a major change, and both the maintainers and the forum/irc members have made reasonable efforts to help when they can. I also look forward to the DSL with 2.6.x project. I just hope that: The base system size is in the same general ballpark as current DSL (50-64megs) so that "toram" can be used in most computers. Some way to do a bootfloppy can be developed. Maybe a 2 floppy disk root/boot LILO or GRUB system instead of a single floppy SYSLINUX system? I wish you the best of luck with the effort. Posted by John on Jan. 17 2006,18:28
Chris, you are taking this thread too far off topic, I deleted your last post. Thank you Cbagger01 and the rest who have given constructive and on topic feedback. I am going to lock down this thread now, the relevant areas have been discussed.
|