There are no longer any paid Club memberships (we abolished that system last year). There's only the Powerpack subscription, which gives you no privileges, it's a simple product which gives you access to the Powerpack edition for 12 months; nothing else.
The only software that is exclusive to the Powerpack edition is software that is not only non-free, but commercial: that is, software we *could not legally include* in any edition that's free to the general public. Software that we have to pay a license fee to the owners to include in Powerpack, and that they only let us license on a limited basis for paying customers.
All publicly redistributable non-free software in Mandriva is available in the public non-free repository, completely free of charge, to anyone. When you set up 2009, this repository is configured automatically.
To put it another way: everything you get for free with Ubuntu, you get for free with Mandriva. The only stuff you are paying for in Powerpack is stuff you could not get for free from any other distribution.
We provide the latest code X.org has for all drivers, and the hardware detection database is maintained. Aside from that there isn't an awful lot we can do to help older neglected drivers - even if we had the time to devote to maintaining ancient X drivers (which we mostly don't), we don't have the hardware around to test. But we do what we can.
There is no x86-64 One. Really, for most workloads, you may as well run i586 on x86-64 systems. If you really want a native edition, install x86-64 Free.
It's not really 'passing the blame'. KMilo is the bit of KDE which handles multimedia keys. If it doesn't exist for KDE 4 yet then, well, we just can't really handle multimedia keys in KDE 4. There's no way around it.
We can do a partial hack - bind the volume up, down and mute keys to kmix via KDE's own hotkeys support - and I'm trying to get that done as an update.
The question was whether issues on that level should stop us using KDE 4 for this release, bearing in mind the release will be the current one for 6 months, has a life cycle of 18 months, and KDE 3 development is entirely dead upstream. We decided that, despite there being niggling issues on that level, it still made more sense to make the jump.
The suspend-on-lid-close issue is a known bug in kpowersave in KDE 4, it's documented (with a suggested workaround) in the Errata.
Basically, One is a hybrid live/install CD which includes proprietary drivers and browser plugins. Free is a traditional installer edition (2xCD or DVD) which is 100% free / open source software, no NVIDIA / ATI drivers or anything (though you can add them from the non-free repository after install, if you're that way inclined).
Relative values are virtually useless without the accompanying absolute ones.
A 0.8% return rate is four times higher than a 0.2% one - but neither is worth losing any sleep over.
Anyone know the actual rates involved, here?
As someone posted already, it's VANOC - the committee formed specifically to manage the Vancouver games - that's filed the trademark, not the IOC.
They've also been trying to force a local Greek restaurant called Olympia (which has been around for 20 years or so, IIRC) to change its name for a couple of years now. They're not getting anywhere with that...
That IBM thing is just about parallel boot, which as I've already explained in another comment, Mandriva has been doing since 2006. Ubuntu and Fedora, at least, also have parallel boot implemented. I'm not sure about SUSE. Doing it via make is a hack job, and no-one actually does it that way these days.
Er, no. You're talking about application packaging dependencies, which have no relevance here. The issue here is what services depend on other services at run time. The method for codifying that *is* an fd.o standard. This is how an fd.o standard header for an initscript looks:
### BEGIN INIT INFO # Provides: $remote_fs # Required-Start: dkms $network # Required-Stop: dkms $network # Default-Start: 3 4 5 # Short-Description: Init script for Filesystem in Userspace. # Description: Init script for Filesystem in Userspace. ### END INIT INFO
The Provides, Required-Start and Required-Stop lines are the dependency stuff.
...because it's booting a much wider range of services appropriate for a generic distribution installation, not (as you are) an extremely minimal set of services entirely tailored to your use.
Do you really think you could release your exact configuration to the general public for use by everyone, and it would work fine?
It's very easy to get a fast boot on a specific system for a certain individual's needs. It doesn't bear any relation to optimizing the generic case.
These are completely different types of work. What Arjan is doing is tailoring boot to a specific set of software running on a specific set of hardware, using an entirely legacy-free init system.
This is nothing at all like what Fred is doing, which is optimizing a legacy boot system for completely generic hardware and software - it has to run on any system, with any set of software available from the Mandriva repositories installed.
The two types of work are utterly and entirely different.
For the record, another of our engineers - Claudio Matsuoka - has been working on the *other* type of boot system for several months now. It began as a re-implementation of the 'fastinit' system found in the Xandros distribution on the Eee. This system is called 'finit', and you can find it at http://helllabs.org/finit/ . It is used in Mandriva Mini, our custom edition for netbook OEMs. It pre-dates Arjan's work substantially, or at least the public announcement of it.
Not to mention that the summary talks about Comcast and then cites an estimated cost based on Time Warner's far more restrictive scheme (40GB cap plus $1 per GB). Talk about bait and switch.
As you correctly point out, capped providers regularly increase their caps in any case, so the projection is entirely worthless. My provider, Shaw, has had caps for a while. They are increased periodically. Most recently, for instance, the cap on the regular service was increased from 20GB to 50GB, and the cap on the premium service from 50GB to 100GB.
As Fred's post mentions, Mandriva has been using a parallel init system (called prcsys) since January 2006. It's entirely SysV compatible and requires only that the SysV init scripts have dependency information (in fd.o standardized format) in the headers. It transparently handles init scripts with no dependency information (they're started serially, after everything else) and can be disabled with a single kernel parameter, 'nopinit'.
It pre-dates upstart, which only showed up in late 2006.
Unless your box is actually doing something useful overnight, you're wasting a lot of power leaving it switched on all the time. Never mind the tree-hugger angle, that's costing you money.
If you look at it in a logical way - how much of the time spent using a computer is spent booting - no, it's not important at all. Optimizing almost anything else is more efficient. However, people aren't logical. People tend to place more emphasis on time spent waiting for the system to boot than a lot of other bottlenecks. You can argue about the reasons for this, it's an interesting psychological debate, but the fact is that it's true. Boot speed has a big psychological effect on how fast people think a system is. So, it's a good idea to optimize it.
(I work for MDV, for those who don't know. I have a small supporting role in Fred's post, credited as 'other colleagues'.:>)
"Formula one racers are fantastic machines, but one could hardly call them daily drivers. There are a lot of things you can do with a motor and a car if you don't have to worry about it lasting 150,000 miles or 8-10 years or driving in the snow or sub-zero weather, sandy or salty environments, or being reliable outside an oval track."
Tiny nitpick - F1 cars don't run on oval tracks. That would be Indy cars. But you're right. Besides, F1 cars are only 2 litres not because it's the best way to build a fast F1 car, but because of regulations. F1 cars would probably go one and a half times as fast as they do now (and be absolutely psychopathically dangerous, not to mention even more insanely expensive) if there weren't a lot of regulations to limit how fast they're allowed to make them.
Actually, newspapers print extracts from speeches politicians haven't yet given all the time - because they're scripted days in advance and, these days, intentionally provided to the press.
If you look at the 'dialog' in the story it's really nothing of the sort. It's canned phrases. Just like, as someone already noted, "One small step..." So there's obviously some phrases that it is known will be used during a successful launch, so the journalist can take a short cut in writing a story about it...
All major news outlets have pre-written obituaries for just about every major celebrity who's vaguely within range of snuffing it. Every few months there's a 'controversy' when one of 'em gets mistakenly published. I don't really see any difference to that...
This can also affect Mandriva Linux 2009 pre-releases. To be clear, the bug is in the upstream kernel itself, not in any code specific to any distribution. It affects any 2.6.27rc kernel, whether it's in a distribution or a clean upstream build.
We have posted a full, detailed notification of the issue for Mandriva users.
2008 Spring was in fact the first major distribution to be specifically designed to work with the Eee (701) out of the box. Mandriva was ahead of virtually every other distro.
Which, to note for the record, is a project of a Mandriva developer.:) If you read westyvw's link you'll know, but to explain it in this thread - netbook customized distros do not achieve fast boot by cutting services from a traditional SysV init, that doesn't get you anywhere enough gain. They use completely customized fast initialization systems. finit is a re-implementation of the one found in the Xandros distro on the Eee; Claudio Matsuoka, one of our developers, first reimplemented fastinit and is now working to improve it and make it more generic so it can be used on multiple systems rather than being tailored specifically to the Eee.
Thanks for the support, Mark.:) The Slashdot story actually missed the most important thing about Mandriva Mini - there isn't an ISO. Well, not for you to download, anyway. This is an offering intended for OEMs, not end users. We want to sell Mini to system manufacturers and redistributors to be the pre-installed distro on the system. We're not targeting it directly at end users. The idea for end users is still that if you as an end user want to install Mandriva on your netbook, use the main distribution, as you suggested you thought was the best idea. We continue to set a high priority on customising the kernel and MDV tools in the main line distribution to work well with the most popular netbooks (2009 should work well out of the box on all currently available Eee models, the Aspire One, and the MSI Wind, to name a few).
Slashdot Mirror
Mark is not actually correct.
There are no longer any paid Club memberships (we abolished that system last year). There's only the Powerpack subscription, which gives you no privileges, it's a simple product which gives you access to the Powerpack edition for 12 months; nothing else.
The only software that is exclusive to the Powerpack edition is software that is not only non-free, but commercial: that is, software we *could not legally include* in any edition that's free to the general public. Software that we have to pay a license fee to the owners to include in Powerpack, and that they only let us license on a limited basis for paying customers.
All publicly redistributable non-free software in Mandriva is available in the public non-free repository, completely free of charge, to anyone. When you set up 2009, this repository is configured automatically.
To put it another way: everything you get for free with Ubuntu, you get for free with Mandriva. The only stuff you are paying for in Powerpack is stuff you could not get for free from any other distribution.
Actually, we use 177.70, but we apply the InitialPixmapPlacement and GlyphCache tweaks by default.
I'm not sure how it could take you weeks - SDL is just one source RPM, so all the patches applied to it are right there...
http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages/cooker/SDL12/current/SOURCES/
there's two Pulse patches.
We provide the latest code X.org has for all drivers, and the hardware detection database is maintained. Aside from that there isn't an awful lot we can do to help older neglected drivers - even if we had the time to devote to maintaining ancient X drivers (which we mostly don't), we don't have the hardware around to test. But we do what we can.
Via VirtualBox or VMware, yes. Yes it does. :) And, of course, vice versa.
There is no x86-64 One. Really, for most workloads, you may as well run i586 on x86-64 systems. If you really want a native edition, install x86-64 Free.
You might want to look at the Reviewer's Guide: http://wiki.mandriva.com/en/2009.0_Reviewers_Guide
it contains a lot of that kind of information.
It's not really 'passing the blame'. KMilo is the bit of KDE which handles multimedia keys. If it doesn't exist for KDE 4 yet then, well, we just can't really handle multimedia keys in KDE 4. There's no way around it.
We can do a partial hack - bind the volume up, down and mute keys to kmix via KDE's own hotkeys support - and I'm trying to get that done as an update.
The question was whether issues on that level should stop us using KDE 4 for this release, bearing in mind the release will be the current one for 6 months, has a life cycle of 18 months, and KDE 3 development is entirely dead upstream. We decided that, despite there being niggling issues on that level, it still made more sense to make the jump.
The suspend-on-lid-close issue is a known bug in kpowersave in KDE 4, it's documented (with a suggested workaround) in the Errata.
Shipping with 2.6.27rc8, final 2.6.27 will be provided as an official update when it shows up.
http://wiki.mandriva.com/en/Docs/Choosing_the_right_edition
Basically, One is a hybrid live/install CD which includes proprietary drivers and browser plugins. Free is a traditional installer edition (2xCD or DVD) which is 100% free / open source software, no NVIDIA / ATI drivers or anything (though you can add them from the non-free repository after install, if you're that way inclined).
Relative values are virtually useless without the accompanying absolute ones. A 0.8% return rate is four times higher than a 0.2% one - but neither is worth losing any sleep over. Anyone know the actual rates involved, here?
As someone posted already, it's VANOC - the committee formed specifically to manage the Vancouver games - that's filed the trademark, not the IOC.
They've also been trying to force a local Greek restaurant called Olympia (which has been around for 20 years or so, IIRC) to change its name for a couple of years now. They're not getting anywhere with that...
It means, a general-purpose distro, as opposed to a special-purpose distro tied to a particular piece of hardware, or software configuration.
That IBM thing is just about parallel boot, which as I've already explained in another comment, Mandriva has been doing since 2006. Ubuntu and Fedora, at least, also have parallel boot implemented. I'm not sure about SUSE. Doing it via make is a hack job, and no-one actually does it that way these days.
Er, no. You're talking about application packaging dependencies, which have no relevance here. The issue here is what services depend on other services at run time. The method for codifying that *is* an fd.o standard. This is how an fd.o standard header for an initscript looks:
### BEGIN INIT INFO
# Provides: $remote_fs
# Required-Start: dkms $network
# Required-Stop: dkms $network
# Default-Start: 3 4 5
# Short-Description: Init script for Filesystem in Userspace.
# Description: Init script for Filesystem in Userspace.
### END INIT INFO
The Provides, Required-Start and Required-Stop lines are the dependency stuff.
...because it's booting a much wider range of services appropriate for a generic distribution installation, not (as you are) an extremely minimal set of services entirely tailored to your use.
Do you really think you could release your exact configuration to the general public for use by everyone, and it would work fine?
It's very easy to get a fast boot on a specific system for a certain individual's needs. It doesn't bear any relation to optimizing the generic case.
OK, er, you fail. Epically.
These are completely different types of work. What Arjan is doing is tailoring boot to a specific set of software running on a specific set of hardware, using an entirely legacy-free init system.
This is nothing at all like what Fred is doing, which is optimizing a legacy boot system for completely generic hardware and software - it has to run on any system, with any set of software available from the Mandriva repositories installed.
The two types of work are utterly and entirely different.
For the record, another of our engineers - Claudio Matsuoka - has been working on the *other* type of boot system for several months now. It began as a re-implementation of the 'fastinit' system found in the Xandros distribution on the Eee. This system is called 'finit', and you can find it at http://helllabs.org/finit/ . It is used in Mandriva Mini, our custom edition for netbook OEMs. It pre-dates Arjan's work substantially, or at least the public announcement of it.
Not to mention that the summary talks about Comcast and then cites an estimated cost based on Time Warner's far more restrictive scheme (40GB cap plus $1 per GB). Talk about bait and switch.
As you correctly point out, capped providers regularly increase their caps in any case, so the projection is entirely worthless. My provider, Shaw, has had caps for a while. They are increased periodically. Most recently, for instance, the cap on the regular service was increased from 20GB to 50GB, and the cap on the premium service from 50GB to 100GB.
As Fred's post mentions, Mandriva has been using a parallel init system (called prcsys) since January 2006. It's entirely SysV compatible and requires only that the SysV init scripts have dependency information (in fd.o standardized format) in the headers. It transparently handles init scripts with no dependency information (they're started serially, after everything else) and can be disabled with a single kernel parameter, 'nopinit'. It pre-dates upstart, which only showed up in late 2006.
A lot of people shut down and restart regularly.
Unless your box is actually doing something useful overnight, you're wasting a lot of power leaving it switched on all the time. Never mind the tree-hugger angle, that's costing you money.
If you look at it in a logical way - how much of the time spent using a computer is spent booting - no, it's not important at all. Optimizing almost anything else is more efficient. However, people aren't logical. People tend to place more emphasis on time spent waiting for the system to boot than a lot of other bottlenecks. You can argue about the reasons for this, it's an interesting psychological debate, but the fact is that it's true. Boot speed has a big psychological effect on how fast people think a system is. So, it's a good idea to optimize it.
(I work for MDV, for those who don't know. I have a small supporting role in Fred's post, credited as 'other colleagues'. :>)
"Formula one racers are fantastic machines, but one could hardly call them daily drivers. There are a lot of things you can do with a motor and a car if you don't have to worry about it lasting 150,000 miles or 8-10 years or driving in the snow or sub-zero weather, sandy or salty environments, or being reliable outside an oval track."
Tiny nitpick - F1 cars don't run on oval tracks. That would be Indy cars. But you're right. Besides, F1 cars are only 2 litres not because it's the best way to build a fast F1 car, but because of regulations. F1 cars would probably go one and a half times as fast as they do now (and be absolutely psychopathically dangerous, not to mention even more insanely expensive) if there weren't a lot of regulations to limit how fast they're allowed to make them.
Actually, newspapers print extracts from speeches politicians haven't yet given all the time - because they're scripted days in advance and, these days, intentionally provided to the press.
If you look at the 'dialog' in the story it's really nothing of the sort. It's canned phrases. Just like, as someone already noted, "One small step..." So there's obviously some phrases that it is known will be used during a successful launch, so the journalist can take a short cut in writing a story about it...
All major news outlets have pre-written obituaries for just about every major celebrity who's vaguely within range of snuffing it. Every few months there's a 'controversy' when one of 'em gets mistakenly published. I don't really see any difference to that...
This can also affect Mandriva Linux 2009 pre-releases. To be clear, the bug is in the upstream kernel itself, not in any code specific to any distribution.
It affects any 2.6.27rc kernel, whether it's in a distribution or a clean upstream build.
We have posted a full, detailed notification of the issue for Mandriva users.
2008 Spring was in fact the first major distribution to be specifically designed to work with the Eee (701) out of the box. Mandriva was ahead of virtually every other distro.
Which, to note for the record, is a project of a Mandriva developer. :) If you read westyvw's link you'll know, but to explain it in this thread - netbook customized distros do not achieve fast boot by cutting services from a traditional SysV init, that doesn't get you anywhere enough gain. They use completely customized fast initialization systems. finit is a re-implementation of the one found in the Xandros distro on the Eee; Claudio Matsuoka, one of our developers, first reimplemented fastinit and is now working to improve it and make it more generic so it can be used on multiple systems rather than being tailored specifically to the Eee.
Thanks for the support, Mark. :) The Slashdot story actually missed the most important thing about Mandriva Mini - there isn't an ISO. Well, not for you to download, anyway. This is an offering intended for OEMs, not end users. We want to sell Mini to system manufacturers and redistributors to be the pre-installed distro on the system. We're not targeting it directly at end users. The idea for end users is still that if you as an end user want to install Mandriva on your netbook, use the main distribution, as you suggested you thought was the best idea. We continue to set a high priority on customising the kernel and MDV tools in the main line distribution to work well with the most popular netbooks (2009 should work well out of the box on all currently available Eee models, the Aspire One, and the MSI Wind, to name a few).