The Linux Weather Forecast

kwabbles writes "The Linux Foundation launched the Linux Weather Forecast yesterday. It features 'current conditions' for kernel development, a 'short-term forecast,' and a 'long-term forecast.' Now developers and organizations that want to see when certain implementations/fixes are planned can find answers at this informative and handy site."

Comments

Linux Climatologist

[MyLongNickName]

Although I cannot predict day to day fluctuations, I can say with a high degree of certainty that in 1000 years, Linux usage among the population will be around 62% with a 73 percentage point uncertainty. My models are never wrong.

Re:Linux Climatologist

[eln]

Everyone knows the best way to get cheap karma is to post something that looks vaguely on-topic early in the thread. If you show up too late to get one of the first 10 or 15 posts, you reply to the first post. Doesn't matter what the first post says, moderators usually don't look to see if your post has anything to do with what you're replying to. There are plenty of stories around here where you see karma bait posted as a reply to a totally nonsensical first post. It works because, by my estimation, 80% of moderators semi-randomly assign moderations to the first half page of comments and ignore the rest.

As of this writing, the post you replied to has already collected an "Insightful" mod, despite displaying virtually no insight. So you see, this technique really does work.

This has been your first lesson in Karma Whoring 101. Join us next week when we discuss how to turn a "-1, Troll" into a "+5, Insightful" by replying to your own posts as AC.

Partly cloudy...

[Gonrada]

Partly cloudy with a slight chance of kernel panics.

I don't think it means what you think it does.

[gihan_ripper]

I was looking forward to reading reports of stormy weather ahead, or even of bright skies. But this seems be be more of a roadmap, not really a forecast. If someone can read this more deeply and see the analogy, then please enlighten me!

Re:I don't think it means what you think it does.

[MyLongNickName]

If someone can read this more deeply and see the analogy, then please enlighten me!

You see, it's like a car with a banana in its radiator...

Re:I don't think it means what you think it does.

[eln]

I think they call it a "forecast" rather than a "roadmap" because that way it's only an educated guess, and they only have to be as accurate as your local weatherman is at long-range forecasts (that is, not at all).

Re:I don't think it means what you think it does.

[bfields]

I think the word "roadmap" would imply some control over the process--this is what we plan to do next and when we're hoping to do it by, etc.

The nature of the Linux community is such that there's not really anybody with the authority to give such a roadmap. Linus can say "no" to people, but he can't make them work on anything they're not interested in. And it's difficult for the various groups implementing new features to guarantee somebody else won't come up with a valid objection that will delay acceptance.

So "forecast" seems more appropriate. They can examine various projects, guess which are closer to being done and which aren't, but can't promise anything.

Re:I don't think it means what you think it does.

[TeknoHog]

You see, it's like a car with a banana in its radiator...

I recall a MacGyver episode where they patch the bullet holes in a car radiator with a couple of eggs, as that stuff solidifies easily. That, in turn, reminds me of vegan baking where people often replace eggs with bananas. So, it's a perfect vegan MacGyver analogy!

Obligatory George Carlin

[RealProgrammer]

The Linux forecast for tonight is ... dark.

UIO: Userspace IO drivers.

[serviscope_minor]

Wow. Just wow. This is a fantastic development. This is an excellent complement to FUSE (userspace filesystems), libUSB (userspace USB drivers), X11 (many graphics drivers), sound mixing, and all the other random stuff farmed out to userspace where possible.

Having develpoed in both user and kernel space, I know that userspace development is vastly easier: a crash is now only a segfault, debugging can be done easily and the most suitable language for the task can be used. This makes it faster and easier to develop both robust and efficient code.

But, come on and look at Linux: it's slowly becoming a microkernel: everything that can be in userspace is slowly moving there. And that is a good thing.

Re:UIO: Userspace IO drivers.

[serviscope_minor]

I agree. Userspace drivers open the way for non-GPL dirvers. Whether you're idealistic or pragmatic, closed source drivers are not a good thing. Just look at the stability of the Intel graphics drivers, compared to ATI (awful) and Nvidia (quite good, but still can causes crashes).

Sadly, UIO can be abused to make non GPL drivers. However, since there's now a nice socket, reverse engineering should be easier, and also I expect that the OSS benefits will greatly outweigh the few hardware manufacturers with misplaced priorities.

Look at FUSE for the wonderful profusion of interesting filing systems. Now Linux has all those features missing from traditional UNIX. You want versioned filesystems? Choose your VCS. Want a database filesystem? Choose the one which uses your favourite database backend. Want a real (as in works for GUI and commandline programs) recycle/trash bin? Got that too (a feature which is unique to Linux as compared to other mainstream OSs).

And so on. Userspace is wonderful, but can be abused. But it's too wonderful to ignore.

Re:UIO: Userspace IO drivers.

[Ant P.]

Yeah, but at least with a kernel driver you _know_ it's crashed.

If that X server hangs is the display just frozen, has X hard-locked the console, or is the whole thing hosed requiring a hard reset? In times like that you're left with crashing the kernel manually using the reset button because there's no other way out. KGI/GGI was supposed to solve that particular problem, but nobody seems to care.

I've had USB devices hang the machine too, but X11 is a really bad offender.

Re:UIO: Userspace IO drivers.

[serviscope_minor]

You'll know if a userspace driver crashes. The device will become unavailable. Then look for the coredump or logfile. But at least you have a coredump to run gdb on now.

But X doesn't always hand the machine. Often you can ssh in and resrart at. Also, that way, you don't loose files which haven't been committed to disk.

But it's true, userspace drives don't magically make software better. X can crash the machine, especially as it's doing funky stuff over the AGP/PCI[X] bus.

However, X is one of many things. In general less kernel crashes is a good thing. If nothing else, it makes drivers easier to debug, which means less driver crashes, too. Also, C is a great language, but no language is perfect for all things. Some drivers may be much easier (therefore shorter and fewer bugs) to write in other languages. Again, that leads to fewer driver crashes, which again is a good thing.

Re:UIO: Userspace IO drivers.

[trifish]

But, come on and look at Linux: it's slowly becoming a microkernel:

In other words, it's slowly getting towards the superior desktop OS design -- the one that Windows uses. And it's really no wonder.

and for the weekend outlook

[kc2keo]

we will see some segmentation faults and a 30% possibility of a new error called kernel disintegration. Also in international news the war on errors seems pretty grim.

That's great, but...

[xednieht]

Is there a way to summarize the summary page with a couple of graphics or pictures and then offer links to greater detail. That first page is a lot to digest at a glance. Weather.com does a good job summarizing mountains of info, perhaps a similar approach for the summary would help.

Also, the xml feed is has a slight syntax error:

XML Parsing Error: xml declaration not at start of external entity Location: http://www.linux-foundation.org/index.php?title=Li nux_Weather_Forecast&action=history&feed=atom
Line Number 49, Column 1:
<?xml version="1.0" encoding="utf-8"?>
^

That's a flawed analogy.

[RealProgrammer]

The kernel is nothing like a car, with or without a banana in its radiator or anywhere else. I'm sorry, but that just doesn't describe the situation at all.

Imagine, if you will, a perfect state machine with N inputs and G(N) outputs, where each output is a Thorgen-Zeta function of all the inputs bounded by the radial square root of each of its eigenvalues. Clearly, the scope is integrable under N, which is probably what led you to your assumption. But where your car-fruit analogy falls down is in assuming that complete T-Z continuity with respect to time.

So a better analogy would be a car with a fish in its tailpipe, dripping maggots along the highway in the rain. Some of the maggots survive to become features, but some are squashed by schoolbuses full of sweaty cheerleaders.

Re:That's a flawed analogy.

[eln]

Yes, of course you're right, but I think if you reversed the polarity on the graviton fields around the vehicle, thus allowing the boson radiation to properly align with the tachyon pulse beam naturally generated by a properly seated banana, I think you'll find the OP's analogy is actually quite apt.

Re:That's a flawed analogy.

[Adambomb]

The day busloads of sweaty cheerleaders are within a kilometer of anything relating to Linux is the day the adoption rate breaks the gauge.

Re:That's a flawed analogy.

[SaDan]

No kidding! I'd love to adopt a sweaty cheerleader!

Wait... what?

Re:That's a flawed analogy.

[cyphercell]

I take it you don't have video working in Linux yet?

Expect ...

[Midnight Thunder]

If its anything like regular weather forecasts, then expect yesterday's forecast to be better than tomorrow's, and long term forecasts to be totally random ;)

Neato

[LordPhantom]

Cool.... now we can all shelter whenever there is a SEVERE LINUS WARNING.

Linux Foundation

[jshriverWVU]

Not to sound lame, but what is the Linux Foundation? I know gnu fosters development of a lot of software, sourceforge hosts a lot of projects, linus and the huge team does the kernel, Ubuntu/Redhat/Debian/Suse/etc take all of the software to make a distribution. I just dont see what the LF does for the community.

Not flaming, if they provide a good resource more power to them.

Re:Linux Foundation

[NickFortune]

I just dont see what the LF does for the community.

Well, they formed from the merger of the OSDL and the Free Standards Group. So they do everything those two thing groups did, including pay Linus to hack on the kernel.

Looking at their about pahge, they also provide legal services, work to promote standards, and provide a neutral forum for debate. The also run sites like LinuxPrinting.org (as it used to be called).

How much of that is truly useful is perhaps open to question. One of the legal services for instance is protecting the Linux trademark, which hasn't proved the most popular activity in the open source milieu. I've also seen concerns voiced about the joining procedure (you buy your way in) and the fact that the board seemed overly staffed with corporate types, with actual developers being a bit thin on the ground.

I think the big trouble is that everyone knew who the OSDL were. This new entity is going to take a bit of getting used to. That said, they seem to be doing good things, so more power to 'em.

My Forecast

[GooseYArd]

My forecast is this forecast will be maintained for about 3 weeks.

Ah.

[RealProgrammer]

I can see that now. Yes, I had failed to account for the natural tachyon output of properly seated bananas (and the lower but significant output of properly seated plantains, as well). But we may be drifting slightly off-topic.

The OP is correct: the Linux Weather Forecast is like a car with a banana (properly seated) in the radiator.

I'd just like to see fewer regressions...

[Erwos]

Something that drives me particularly nuts about Linux is the number of regressions that happen. Don't get me wrong - I love Linux. But it feels like, in their haste to get things done better than ever, they sometimes reintroduce old flaws.

Case-in-point: Firewire, and more precisely, multi-LUN support. Sounds esoteric, but it's actually not too uncommon to find hardware that needs it these days. When 2.6 came out, 1394 (as a whole) was just plain broken. They finally got it fixed in 2.6.12 or so, but then, in 2.6.22, they introduced a new Firewire stack - which promptly broke multi-LUN support. Maybe not everyone needs this, but I'm heavily dependent on it. I'm now in a position where I can't do any kernel upgrades until I've confirmed the fix has made it into 2.6.23 final.

I know it's something of a petty gripe, but I'd appreciate it if Linus could do a better job of making sure regressions like that didn't slip back in. I don't need a repeat of the bad weather that we've already had. :)

Weather Forecast?

[Your_Mom]

Personally, I'd rather have a "linux-kernel flamewar fire threat gauge"

I'd check that hourly.

Today's LKML Flamewar Threat: "Moderate"

[Cid Highwind]

Low: Peace and harmony, everyone must be on vacation this week
Moderate: Noobs complaining about nVidia driver crashes
Elevated: Linus just tried GNOME... again
High: Con Kolivas and Ingo Molnar go mano-a-mano over scheduling algorithms
Extreme: Hans Reiser is back, and he's armed, dangerous, and off his meds!

Not quite.

[jd]

The seating must be in the Complex domain, and for the output to generate the necessary synchrotron tachyon, the banana must be entangled with the fish in the tailpipe, as postulated by Minsky space.

Re:Like predicting Sunny in California

[Jello B.]

Compiz Fusion. Microsoft was chasing Compiz's taillights with Windows Vista.