, Linux sucks
... It just usually sucks less than other operating systems. Here are a
couple of instances:
- Memory leaks. Linux doesn't handle them very well (neither does
Windows). The folks who wrote k3b, for example: they cleverly programmed a
memory leak into k3b, such that -- under certain circumstances -- k3b will
start eating memory, until the machine is unusable. You can't kill the
process, because Linux can't fork new processes, because the kernel is
spending all of its time swapping. While this is bad, it is worse that Linux
lets it happen, which can drag a machine to its knees. There's a simple
solution: if a process is requesting more memory than a machine has, then
start fucking rejecting new requests for more memory from that process.
- You can't terminate processes. There is no super kill. If there is a
process that I want killed, let me fucking kill it and free the resources it
is using.
- If you have a bad CD -- say, burnt with a corrupted ISO, or some plain
text masquerading as an ISO -- and you then try to read that CD, the kernel
will go into a loop trying to read from that CD, receiving repeated I/O errors
on the device. If a user tries to eject a device, and s/he, say, passes a
"force" flag, then let him/her fucking eject it.
If the kernel is so stupid that the user feels s/he needs to force it
to do the right thing, then let them do it.
Posted at #
GPLv3, DRM, and Han Solo (a response)
Peter Rock posted a blog
defending the GPLv3 in response to an article on CNET by Jonathan Zuck. Jonathan's article voiced numerous concerns regarding some provisions of the GPLv3, and Peter Rock believed this to be an "astonishing article of misinformation".
I think Peter's misinformed about the GPLv3, and here's why:
- He says: "Jonathan states ... prevents hardware companies from controlling the final implementation". He follows this with: "This is not true. GPLv3 allows the owner of a device...". How does allowing the owner of the device to change the final implementation negate Jonathan's statement? Jonathan's statement is still true. The hardware companies don't control the final implementation under the GPLv3.
- He says: "The only instance where this would be true is when the user of the phone is also the owner of the phone." That's a pretty fine nit he's picking there. How often is the user of a cellphone not also the owner of the phone? While in the strictest sense, Jonathan's statement is false, it is a minor point, and Peter's just being argumentative here.
- He says: "Depending on the motive of the regulator, this may be true. But Jonathan makes me feel uneasy in that he actually perceives this as a strike against GPLv3." Oh, hell yeah. Allowing users to reprogram cell chips is a bad thing, no matter how you look at it. A single user can screw up an entire cell network with a badly programmed chip. I'm all for user rights, but this is just a recipe for disaster, and I'm for almost anything that prevents people from gaining the right to dick up my cell reception.
- I agree with Peter that Jonathan misunderstands the whole point of the private key provision.
- On the other hand, I think Peter misunderstands it, too. The GPL requires that the vendor release the private keys for devices. It is true that this key release doesn't have anything to do with the owner, but the owner isn't the one who (for example) went through the FDA process to verify that the device doesn't kill people. The owner didn't write the software that the FAA authorized for use in transporting people. Here's a extreme example that is, nonetheless, plausible: Company A creates a GPL-3 medical device. It spends 6 years going through FDA verification and validation (this isn't an exaggeration). Company B buys said device, gaining access to the private keys (which can be used to sign new instances of software to run on said device), modifies the software. Said device subsequently kills X people. Now, the only reason this device is legal to use on people is because the original company went through the FDA. As soon as the owner modifies it, it is (technically) no longer the same device. So, it could be argue that it is legal to modify the device, but illegal to use the device once modified; the original company is legally off the hook. Or is it? How do they prove that their original software was modified? They had to give up their private key under the GPL-3, so there's no proof that that was not their original code. But let's ignore that -- how can one justify removing a legitimate safety measure (the device will only run FDA-authorized source code) on a medical device?
As a potential patient, as an occasional air traveler, I do not want some Joe Dipshit re-writing the software that's going to drive the medical devices being used on me, being used to keep the airplane I'm riding in up in the air. It is bad enough that I have to trust the original vendors, but at least they had some oversight and verifiable review process. No, keep the GPL-3 off my cars, planes, trains, defibrillators, anaesthesia machines
Posted at #
WMII
WMII-3 is a great window manager, but it is broken. One app, in particular, can crash WMII; that app is IntelliJ, a Java application. This app crashes no other window manager, and no other Java app that I've encountered crashes WMII, so it is definitely some interaction between WMII and IntelliJ.
Granted, IntelliJ could be the app at fault here, but I propose that no application should be able to crash the window manager, no matter how badly, or maliciously, written. Therefore, this is a bug in WMII, and it is (unfortunately) keeping me from using WMII at work, where I have to use IntelliJ.
Still, if you're somebody who uses the keyboard a lot, I recommend WMII -- it keeps your hands on the keyboard
Posted at #
King Abdullah of Jordan
I heard an interview of King Abdullah II of Jordan this morning. Man! I can't listen to the news without getting pissed off these days. Jordan gets King Abdullah -- smart, multilingual, and eloquent... what do we (in the USA) get? A chimp .
I mean no disrespect to Chimpanzees
Posted at #
Eclipse sucks
I almost wrote a blog titled "Eclipse rocks", but now I can't get Eclipse to start up.
No, nothing has changed. I haven't installed any new software or plug-ins. I have 1.5GB of RAM in this machine. I shut Eclipse down a couple of days ago while I was doing some bug resolution verification, and it fails to run today.
WTF? Now, granted, Eclipse may not be at fault. Java has its fair share of bugs, but I don't give a fuck who's to blame -- this is really annoying. And there's no way to track the problem; Java seems to be running, and an strace shows that it is doing something. But Eclipse just sits there and hangs on the splash screen.
The company I work for are not paying me to track down some other developer's bugs, so I have to ditch Eclipse and go back to IntelliJ. Which is no joy to use, but despite all of its bugs, at least it runs.
Mind you, I feel that this is just another example what I feel is fundamentally wrong with many projects and languages being developed these days: there are far too many layers of software building up to the final application. Code reuse is good, but every layer you add is another point of failure. Java just has too many layers. It is an OS, running on another OS. I've never seen Kate, a natively compiled application, exhibit this kind of arbitrary, unpredictable behavior. However, I see this happen to Java apps all the time.
Oh, and I don't use Kate for software development because the language bindings are pretty weak. IntelliJ and Eclipse both have code browsing (class name lookup, method dereferencing, etc) that is absolutely required for working on large projects -- when you're dealing with 3,959 classes and 357,000 lines of code, you really need this. Well, I do
Posted at #