Yes, and here's the catch. It has been my experience that if you want a Linux desktop that can compete with the offerings from Microsoft, in compatibility, functionality, and eye candy, you are looking into something pretty heavyweight: Gnome or KDE, OpenOffice, Firefox, all of them heavyweight stuff, not to mention the ever-increasing number of deamons that are running in the background (hald, NetworkManager, Dbus, ...). It doesn't work well on low-end hardware on which Windows XP will still run smoothly.

We use cfengine with close to 100 machines and works quite fine. My only gripe is that on Ubuntu 8.04, it has a bug such that it can't determine which packages are already installed. And since Desktop is first priority for Ubuntu, their maintenance of software for larger environments is abysmal.

There is no language called POSIX. It's an OS standard, and software authors writing in high-level languages should not need to care about it. There are languages like C, C++, Java, etc. If you're writing Java, you should pay attention to the Java standard. If you're writing C, you should pay attention to the C standard (which doesn't include OS specific things like fsync()).

And if you adhere to the standard of the language you are using, you should have some reasonable expectation that things will work, instead of dying a horrible and gruesome death. A conforming HTML page should not fail. A conforming Java/C/C++/whatever program should not fail either, but the problem is, it will. This is like asking HTML authors to understand the low-level details of the browser rendering engine in order to write working HTML pages.

Some high-level languages (e.g. PHP) have no built-in fsync. Also fsync() is not part of the C standard, it's a POSIX extension. What you have in C is fflush(), but that will not fsync(). So books about programming in C usually don't cover fsync(), as it's not part of the language. I know that sounds like nitpicking, but truth is, if you've learned programming from books about some language, chances are you've never heard about fsync().

Basically, what happens is that you need to understand OS design in order to program in a high-level language, and nobody (at least none of the books) tells you so. This is a WTF on more than one level... either make it part of the language, or make sure it isn't needed.

An enterprise would likely want to use the LTS version (Ubuntu 8.04), but CFEngine is broken in Ubuntu 8.04 (can't figure out whether a package is installed). Don't know about puppet, but we're using Ubuntu on the desktop (70+ machines), and we're somewhat pissed off by the fact that Ubuntu completely neglects QA for enterprise use (issues with NFS-mounted home directories, failure to fix an entirely trivial one-line bug in CFEngine, ...).

See above - get the most recent version of gcc, use 'gcc -O2 -Wall'.

It's splint these days (at least on Linux). And using splint on any nontrivial large code base will bury you in tons of mostly irrelevant warnings. If you dare to attempt cleaning up the mess, you'll find that you have to annotate your code. And then you have the problem that splint will only spit out correct warnings if your annotations are correct, so you have just doubled the potential sources of error (now it's annotations+code instead of just code).

Tools like splint just don't understand the flow of control in a program, which makes them next to useless. Static checking must be implemented in the compiler. Gcc is actually quite good already in that respect, and constantly improving - the version that ships with Ubuntu 8.10 detects the array overflow:

$; gcc -O2 -Wall -c -o bo.o bo.c
bo.c: In function 'f':
bo.c:4: warning: array subscript is above array bounds

As far as I can see the whole "real Unix" thing is primarily about meeting a set of requirements...

Yes. And as a developer, I can say that this means A LOT. It means that software will "just work", without having to care about obscure and stupid incompatibilities, and that's important.

The problem is that Wikipedia encourages the use of secondary sources and discourages the use of primary sources. According to Wikipedia policies, it probably would have been "original research", and thus unacceptable, if an article author would have tried to get hold of the primary source (copy of birth certificate).

The problem is that wikipedia itself discourages the use of primary sources by wikipedia authors. The reason they give is that using primary sources may lead to "original research", which is strongly discouraged as well.

That's what you get if you discourage the use of primary sources in favor of secondary sources.

My point was that I'd expect sick Mondays to be much more prevalent than sick Fridays (more than any other weekday even), and for a good reason.

Well, spending the weekend with the kids is a great way to get infected with whatever they bring home from kindergarten or school. There is a good reason for sick Mondays.

While open source software generally sucks big when it comes to privacy protection*, there's actually an Apache module to replace the connecting IP with localhost. Which means neither the logs, nor any web app running on the server will ever see the real IP adress.

*Yes, it does.. eliminating privacy violating information like IP adresses, as even required sometimes by law, is difficult, or even next to impossible with default software from your preferred Linux distro. I'm trying to run a server on a 'we log strictly anonymously' basis, and I haven't fully succeeded yet.

Uhh.. pardon me, who moderated this insightful?

All low-mass stars, including the Sun and ranging up to F-Type stars (about 1.7 times solar mass) have an outer convection zone (meaning that outside the core, and up to the surface, energy is transported by convection). There's nothing 'bursty' in that mechanism.

Hard radiation of low-mass stars is generated in the corona, which is heated by magnetic reconnection events (the magnetic equivalent of a short), leading to sudden release of the energy stored in magnetic fields. This is what is called 'stellar activity': starspots, flares, X-ray radiation, ...

Some red dwarfs are indeed much more active than the Sun, many are not. Activity is generally connected to the age of a star since magnetic fields are generated by a dynamo mechanism, and stars spin down slowly as they are aging, leading to a less efficient dynamo and a decrease of activity.