Eclipse Fall Down Go Boom

September 28, 2012

EclipseCrashAgain

Blows up before it even runs, looks like the Mercurial plugin having a bad day.

No point bothering to report the bug, it’s a pain in the ass and they never seem to fix them. Just for example, notice how the text in the picture naming the file above looks like it’s selected, so I could copy and paste it when opening an editor to look at the contents of the file? That’s a lie, like a “pull” handle on a “push” door, and it’s long-ago-reported bug, never fixed.

I’ve been looking for an excuse to go try IntelliJ IDEA, and this just might be it. Read the rest of this entry »

User experience, starting a few weeks ago:

Using the Little Snitch firewall, a program called “ksfetch” wakes up every hour or so, asking for permission to connect to the world.
I check to see that it is Google software update related, decide that is okay, and give it eternal permission.
A little while later, even though it has eternal permission, I am bothered again. And again, and again, and again, and again, and again.
WTF?

What’s happening is that the Google Software updater (“Keystone”, apparently) frequently writes out a new version of ksfetch, and because it is a new application, firewalls like Little Snitch decide that it has not been approved for access. Why this fails security ergonomics is that it is very annoying, and it trains people into the bad habit of automatically approving access by anything named “ksfetch”; if I were writing a virus, I’d be sure to call it “ksfetch”.

You can see people attempting to figure this out on a Google product forum here. There’s one reply from someone who might be with Google, but they’re clearly not getting it.

New laptop glitches

June 18, 2012

This is more or less a log, maybe I’ll bother Apple about it. The from-backup-disk install (Lion to Lion) was surprisingly good, but there were a few glitches:

  • The command line tools for XCode were installed on the old system, but not the new.
  • Java was installed/enabled on the old system, but not the new.

This was a surprising pair of omissions, because (on the other hand) all my MacPorts stuff woke up happy, including the auto-started apache daemon for the Trac server I run on my laptop.

Whoops, one more glitch — it did not automatically transfer over my iTunes authorizations, or ask me if I would like that to happen. And for some reason, the store is not answering the request to authorize the new computer. DRM, I hate it. This is why I don’t buy DRM’d music anymore — not sure if iTunes store is or isn’t nowadays, because when I switched to eMusic, it was, and Apple hasn’t fixed the problem yet for the music I bought from them so long ago. (Dear RIAA: I curse you guys everytime this happens. Way to screw over your paying customers; the pirates deliver a superior product.)

Happiest surprise: realizing that with spaces and swipes (no idea how long these have been supported, but only with Lion did I start to pay attention), I could run my “Complete” New Yorker in its own screen-hogging “space”, and just swipe back and forth to it at my convenience. Next thing to try is to see if this same trick works for screen-hogging games like Civ 5; I fear it will not.

Consider this a bug report. Workaround is to disable iCal with extreme prejudice, and get BusyCal.

Read the rest of this entry »

Trying to jump to a source line from someplace in the stack trace, it fails with this message instead:

ArchiveDoesNotExist1

“Archive does not exist: /Users/dr2chase/worksapce/PFC/default_repository/caches/bytecode_cache/*”

I had to type that in by hand because the Eclipse guys have not yet mastered the art of rendering selectable text that I can copy and paste in their little alert windows. (That’s another bug.)

Of course, the archive does exist; that character string is the convention used by another part of Eclipse to designate “all the jar files in this directory”.

ArchiveDoesNotExist2

Their error message is doubly bogus, because the line I am clicking corresponds to a file that appears somewhere else on the debug classpath. It could simply skip the error and carry on to see if a later entry succeeds. A likely workaround would be to reorder the entries in my classpath, but that is not an option here because we have duplicated class files on the path, and order matters.

This is not a showstopper by any means, but it is tedious, and removes a feature from Eclipse that I normally use and expect to work.

Eclipse version:

Version: Helios Service Release 2
Build id: 20110218-0911

MacOS version 10.6.8

And why I am blogging this, instead of entering it into their bug database? I’ve tried that in the past; they yield equally satisfactory results, but this is easier for me.

Just tried one. The online instructions (three or four different versions) are insufficiently detailed, and the first step is apparently impossible. It will only waste your time and money, and maybe break your iPod. The worst instructions tell you to use a “safe opening tool” (with link to a bunch of spudgers) but show a picture of a razor knife. My guess is, if they told you to use the tool that worked (the knife) and you cut yourself, they’d get in trouble, so they tell you to use useless spudgers, but show you pictures of knives so you’ll eventually try that (but the arterial bleeding will be your fault, not theirs).

So if you are tempted by one of those products online, Just Say No. Get a professional to repair it instead. Learn from my mistakes.

It’s better, but its behavior when adding a new 2TB disk drive is classic Unix-stupid.

I ran the fancy-GUI program for formatting the drive, everything is looking good, I click “GO” and it proceeds to tell me (as it formats the drive) that I chose my partition alignment poorly, and performance will suck. “Repartitioning is recommended”. Oddly enough, merely repartioning the disk a second time (which is what I was doing in the first place, RIGHT?), DOES NOT MAKE IT BETTER. Read the rest of this entry »

No idea. Googling didn’t give me good answers. Read the rest of this entry »

I’m looking into trying Amahi. It requires Fedora 12, I plan to try an install on an old box that I don’t think has a DVD drive. So, download some CD images, burn them, and away we go.

Except…. Read the rest of this entry »

FIXED IN http://download.scala-ide.org/nightly-update-wip-experiment-trunk (This is a nightly build, based on latest as 2011-02-22, Scala 2.9.)
Unfortunately, this version of the plugin takes 7 times longer (45-50 minutes) to perform a from-scratch project rebuild, and incremental changes are just as costly as clean rebuilds. This is clearly not usable. However, it is possible to use the new version, to clear the deadlocking state, and revert to 2.8.0, which can perform incremental rebuilds in a few seconds.

Sometimes, when I start Eclipse (Helios, M20100909-0800) with the Scala 2.8.0 Helios plugin, it never makes it past the splash screen, and CPU consumption falls to zero. Updating to the 2.8.1 version of the plugin is not an option (yet) because of this other bug. When (apparent) deadlock happens, jconsole can be very helpful in determining what’s really gloing on: Read the rest of this entry »

Follow

Get every new post delivered to your Inbox.

Join 137 other followers