Mandrakelinux 10.1 alpha 1 has been uploaded to mirrors. According to the revised schedule estimation, this is a cooker snapshot, coming up with kernel 2.6.8-rc1, gcc-3.4.1, kde-3.2.3, mozilla-1.7.
Mandrakelinux 10.1 alpha 1 Released
2004-07-17 Mandriva, Mandrake, Lycoris 38 Comments
For those who want even more details:
I really like mandrake. It’s easy to setup and just fast. At least way faster than fedora on the same hardware (in case of bootup, loading apps, responsiveness). Drakconf is really good. A little bit weired that, although kde is their primary desktop environment, drakconf is a gtk program as long as I know.
Unfortunately as a gnome user, you don’t get the latest and greatest with Mandrake (instead of Fedora which will ship with gnome 2.8). If they would ship the distribution let’s say a month later the new gnome version could maybe be included…
Also, I really find it a pity that Mandrake won’t ship the new xorg version (probably released on august the 25th.)
Keep up the good work!
Have they fixed the partitioning problem? Wiping out the partition table thing?
that bug was a problem with bios, windows and some versions of the 2.6 kernel, it was not a pure mdk problem…
basicly the kernel overruled the bios about what kind of disk was hooked up (the kernel was from what i understand correct, so it cant realy be labeld as a bug) and wrote a matching partition table. windows, getting its info from the bios and trusting it ended up makeing no sense of the partition table what so ever…
Just tested it, but came back to my trusty 10.0 setup to write this post.
For some reason I couldn’t quite figure out, I had to install it twice to get it right. Then everything worked fine, but there must still be a lot of debugging code because it felt slower than my (by now heavily customized) 10.0 setup.
Anyways, for its first Alpha 10.1 is looking very good indeed.
@ Roberto: I am not sure what you mean. I haven’t ever had Mandrakelinux wiping out partition tables, nor have I heard of that happening with any official release of any one of the “big” Linux distributions. But partitioning tools are never idiot-proof anyways…
If you use Ext2 file system, then you will not run into the partition problem. Of course it may not be what you want to do, just a suggestion.
“Have they fixed the partitioning problem? Wiping out the partition table thing?”
doesnt wipe out anything. just tries to “correct” it which windows doesnt like and it fails. its possible to bypass it b setting hard disk access to lba or recover by other methods. not sure if that has been entirely fixed
its not a kernel problem since kernel doesnt manage partitions anymore. its a parted or whatever partition manager is being used
Does anyone know if they got the SATA support fixed? Last time I tried 10.0 I couldn’t get the boot loader to install because of a bug where it still saw two drives and didn’t know which one to install the boot loader on and would freeze.
“its not a kernel problem since kernel doesnt manage partitions anymore. its a parted or whatever partition manager is being used”
Do the parted people know about this? I know it seems like a silly question but I haven’t heard any updates…
“Do the parted people know about this? I know it seems like a silly question but I haven’t heard any updates…”
no idea.seems to be a complex problem
hey Arthur, thanks for your kind response and for the suggestion! I appreciate it. I don’t really care what file system I use as long as it is not FAT32 ext2 will do just fine
But partitioning tools are never idiot-proof anyways…
Andrew, I am not an “idiot”. Perhaps you haven’t been following the reviews, if you are not aware of this problem. It doesn’t happen to “every” machine’s configurations. It happens only to some. I just can’t be bothered playing with rescue disks if this problem does occur.
urpmi upgrade possible ?
I can’t test yet coz I am stuck in work, so if anyone tries it, would you post the addresses here please ?
I wasn’t suggesting you are an idiot, it’s just that accidents happen a lot with partitioning tools. And there is no way around it: if you include a partitioning tool (even one as safe as Mandrake’s) in a Linux distribution, you are bound to get some people losing some data – and that includes myself, more than once I confess.
Nowadays I just leave a 10GB partition aside for testing new Linux distributions.
Generally it’s best to wait till the kernel, KDE, Gnome, X and MDK-Control panel go stable or near stable. Otherwise it is very easy to render your working MDK system useless if you are not a person with debugger skills.
i tryed to explain it in my other post but i feel the need to expand on it. when windows boots it takes all its clues from the bios, it never bothers to probe the drives on its own.
most bioses have a autodetect system that tryes to guess what kind of disk you have and if it need to apply lba or not. sometimes it gets that wrong and tells windows bad info. windows uses this info to set up and read the partition table.
linux now pokes the drive directly. then it passes this info on to the tools run. therefor the problem isnt so mutch in one area but that diffrent tools comes up with diffrent results. and we all know what happens when 2 or more sources says diffrent things about the same topic?
so what we end up with is that linux pokes the disk, hands the info over to parted, parted finds that the current table dont match the physical drive and fixes it. in the prosess it makes the partiton table unreadable for windows as it rely on faulty info.
in many ways they have the same problem with acpi, only that here the bios is often made to conform to a faulty implementation in windows and therefor linux coders have to introduce “bugs” into the code to get it to work properly.
so in the end the question is, should the linux coders fix the error, knowing that they yet again conform to problematic behavior from windows or just point out the error in the current system and state that its not a linux problem and that the bios makers and maybe windows is what needs to be fixed? personaly i think the first option is like bowing for the emperor even tho you know he have it all wrong…
>acpi, only that here the bios is often made to conform to a faulty implementation in windows
You have it other way around. Windows had to conform to first faulty implementations of acpi standard in some bios-es.
Of course, Windows team could make a stand, say “screw all of you, here is standard, here is your stupid bios, we are doing nothing until hardware vendors fix their mess.”
But they didn’t want to confuse and alienate end users who don’t know the difference between bios and bias, so they did implement acpi the way hardware told them to. ACPI works in Windows, right? What else do you want?
It is not about whose ego is larger and who is the boss, it is about people using or not using your software.
Windows team went as far as “fixing” windows to make badly written but popular applications work on it. You know, when application tries to access memory it just released- it is bad, speaking politely.
You can make a stand and tell everyone go to /dev/null.
Instead, they actually put a workaround in OS code, to hold off reuse of released memory in some cases.
>not a linux problem and that the bios makers and maybe windows is what needs to be fixed.
Yes, exactly that stand. It will take you where? Nowhere.
This is a small but important example of the difference between developers coding software for users, and developers coding software for themselves.
Of course, when user oriented company becomes a monopoly it can afford to ignore users, but then people start to move away from it.
Regardles of how many times you say “it is not the Linux problem” there always be perception that it is and, more importantly, that “Linux people” did nothing to fix it or implement a workaround.
It is lack of care about non-technical end users trying to use your product. Don’t worry about it if you don’t need to win non-technical user base.
so in the end the question is, should the linux coders fix the error, knowing that they yet again conform to problematic behavior from windows or just point out the error in the current system and state that its not a linux problem and that the bios makers and maybe windows is what needs to be fixed?
If Linux is so good it can guess the right partition layout, it has no business ‘correcting’ a partition table other OSes rely on. Were Windows to ‘correct’ something that Linux depended on, how loud would be the outcry? I for one would shout. I just hate it when some piece of software changes something that was there before them and may be in use by others. Oh, I know software does that all the time – it’s just that the bigger the damage, the angrier one gets.
Windows at least has the politeness (the nerve?) of telling you it will change things that may render other OSes unbootable.
“Windows at least has the politeness (the nerve?) of telling you it will change things that may render other OSes unbootable.”
Untrue. Windows doesn’t ask before overwriting the existing bootloader…
I just can’t believe this benchmark. SuSE faster than Mandrake? Not in my experience, where SuSE 9.1 was quite sluggish compared to MDK 10.0, which feels pretty fast. Debian should also fast, but Linspire was not that snappy on my box. Guess it all depends on the hardware you are using and what apps you use most.
I decided to try out the alpha 10.1 to see what is new. I guess I’ll have to wait. It would install but I could not get X to work.
This is a machine I use for testing that has run every version of Mandrake since 8. I also had trouble with the X configuration on the Yoper that I tried the other day. However, Xandros OC just worked fine.
I guess I will dust off my bugzilla account……………
Windows had to conform to first faulty implementations of acpi standard in some bios-es.
Actually, linux developers tend to do the same – to code workarounds for specific ACPI and USB implementations which do not copmply a standard but work with windows.
I wouldn’t trust that benchmark any further than I can type cast it into void.
First of all I don’t think it is very well set up. Install time? Are we talking seconds/minutes/hours? Probably minutes but these things need to be present in order to have a trustworthy benchmark.
Second, I never trust a benchmark when it comes from non-3rd party places. Yoper scores mighty fine in this benchmark. No wonder! This benchmark is hosted at Yoper’s website and was done by a Jon Williams which apparently is the “Yoper business development coordinator”.
This is just like when MS and IBM announces another victorious benchmark achievement on their SQL servers…not really worth anything except marketing bravada.
Just my opinion, others are entitled to theirs
That’s ok Andrew, no problem
Also a msg for Linux developers. “Russian Guy” has a good point and you should make a work-around. By saying “It’s not a linux problem” you will get no where, do you really expect MS to fix this problem so it makes Linux easier to be installed? Also, this problem did not exist before so I am sure you can go back to the previous method of doing things that actually works and I mean this in the nicest possible way. MandrakeSoft, Red Hat did it before, they can do it again
nah – I do not believe those benchmarks either. Yoper is fast…. but fast only in comparison to rh9 on which it is based, it is nowhere near as fast as mandrake or debian…
Also the benchmarks themselves are extremely flawed, it says lower user memory is better ????? since when ?
take note that the benchmarks are from yopers own site.
have a look around for independent benchmarks, and if there is none, then create them
Isn’t ext3 standard nowadays? Didn’t think anybody used ext2 anymore…
(sigh) yep, Mandrake 6 was my first ever distro. I installed it onto an old dell poweredge (had dual 200-something processors IIRC). I set up samba, mounted my windows share, and managed to use it productively at work for a good 3-4 months until the scsi controller died. But now, I’m all about slack-type distros–they just seem cleaner and simpler. I’m just not interested in the big/slick distros.
“Yes, exactly that stand. It will take you where? Nowhere.”
i hope it will put the blame where it is supposed to be and maybe give us better standard compilance in the end instead of letting corps belive that they can shaft standard left, right and center when it fits their bottom line…
just look at the cd-rom “problem” not to long ago, it was the hardware that was useing a bruner signal to initiate firmware updateing that was the problem. the kernel in mandrake just happend to use the same command to see if the device at the other end was a burner or just a reader. if they had followed the standard then there would be no problem, the reader would just return a error message (or not respond at all), not go into firmware update mode…
as for the software, well boho. what they should have done if they wanted to was to create a sandbox enviroment and tell people that if the software broke in a normal run then they should run it in sandbox. if that broke send us this and that output file so that we can update hte sandbox for future use. hardcodeing third party software fixes into the os is just plain nuts!
i would rather have a os that is rock solid then one that have to take into account bad codeing habbits and errors.
this is the power of open source, if one does a upgrade and something breaks one still have access to the code so that one can pay someone to fix it or fix it oneself.
as for the “problem” of haveing one linux overwrite the boot info of a older install. they should add a option in the bootloader config on install that either looked in the MBR for a existing one or give you a dialog that allows you to point out a existing config file on a existing partition that they can add their boot info to. hmm, maybe i should put up a request for this on the mandrake system.
and windows boot is definetly funny. it shafts the mbr without asking, and then moves on to check a file. that file is allso read when you do a install but not allways properly updated. i have done reinstalls where i have tossed out all existing partitions but still the windows loader insists that there are more then one windows os installed on the system!
Isn’t ext3 standard nowadays? Didn’t think anybody used ext2 anymore…
I believe ext3 became a defacto standard because RH (and eventually Debian) switched to it, but now XFS seems to be available on just about every major distro.
I really hope Mandrake does something to improve the quality of this release. I know it has been an ongoing issue for this company for years. Yet, even in 10.0, a freshly installed & freshly updated Mandrake 10.0 system gives an error window every single time a KDE app is exited. Of course, the first guess is always a hardware problem, but given that it doesn’t recur in any installed distro except for Mandrake 10.0, one can only assume it is a Mandrake QA issue yet again. Maybe someday.
Nonsense.. There are no such errors, at least not commonly. Either you hav bad hardware or a poorly configured graphical driver.
The only time I have seen the error that you talk about is when a user didn’t install the Nvidia card proprietary drivers properly. Are you using an NVidia card? If you are, uninstall the drives (read the Readme provided by NVidia on how to do this) and then reinstall the latest driver.
Not as much as your reply.
“There are no such errors, at least not commonly.”
So I guess that means the 100% reproducible error does not exist?
“Either you hav bad hardware”
Your right. Mandrake is the only installed distro with the problem, so it must be my hardware.
“or a poorly configured graphical driver.”
A Mandrake autodetected & autosetup nv driver you mean?
“The only time I have seen the error that you talk about is when a user didn’t install the Nvidia card proprietary drivers properly. Are you using an NVidia card? If you are, uninstall the drives (read the Readme provided by NVidia on how to do this) and then reinstall the latest driver.”
Moot. The Nvidia drivers are not installed.
Look, if you want things fixed, report a bug to Mandrake. Be specific, provide error messages.
I am responsible for 40 MDK Linux Desktops and MDK has been rock solid. We tested Fedora 2, Suse 9.1, Mepis and Slackware 10 before finally deciding to settle for Mandrake. Whatever your problems, they are not widespread and crying about them here doesn’t help anybody.
Mandrake is one of the few companies that despite having a completely open development process and a completely free software distribution always attracts all kinds of trolls.
“Mandrake is one of the few companies that despite having a completely open development process and a completely free software distribution always attracts all kinds of trolls.”
You are the one trying to pass off a Mandrake bug as a hardware problem, so whose the troll?
Where’s the link to Mandrake’s bugzilla with your bug report? How do you expect MDK developers to fix a bug that they are not aware of? Where are the specific error messages? I am not saying that you are not having an issue, I am saying that your whining here is helping no one and that it is fairly common for astroturfers to attack Mandrake.
Mandrake did not work as well as it does, we would not use it, period.
I agree totally with Eu. If you have an issue, send it to Mandrake and let them correct it.
I think the actual problem is that Anonymous is used to reporting faults with his Windows system and nothing being done about them, that in the end he gave up reporting faults and just yapped about it online…
Even just doing a simple thing like a bug report will let the “casual” linux user find out the differences between the Linux world and the Windows world….
I totally agree with raver31 here, and I would even add one piece of information that Anonymous should find interesting: you don’t even need to be a registered user of Mandrakelinux to send them bug reports.
I mean, even if you downloaded Mandrakelinux for free and installed it on 100 machines, if you report a bug in the proper way (yes, it does take a few minutes of your time on the Internet), they’ll fix it.
It’s clear that reporting it here is not going to make it go away, but reporting it through the Mandrakesoft bugzilla website would be the first step.
Regarding the speed of 10.0, I would say it’s one of the fastest distributions around. Mandrakesoft takes the care to build each package with individually customized optimization flags, and their kernel is also tuned for speed. Also, both KDE and gcc have improved significantly in the last few years.
And 10.1, using gcc 3.4.X, will add an extra 7.5% speed to C++ apps. So speed-wise, it’s looking good too.
to the earlier poster about debug code – the big difference is KDE. KDE binaries in non-stable MDK are built with debugging symbols, in stable MDK they’re stripped. Don’t know what speed difference that makes, but it makes a big size difference.
And for another earlier poster – MDK has no “primary” desktop. GNOME and KDE have equal standing. The drak* tools are GTK+ basically because they always have been – a lot of MDK code is written in perl and for a long time perl-gtk bindings were a lot better than perl-qt.