Some bizarro stuff going on with WordTwit. First it was complaining about a foreach() loop and coughing up a PHP error. Now it doesn’t seem to know how to handle the multi-dimensional array $tweets; it spits back the first character of the first levels elements. I posted the letter “t” -900 minutes ago and the number “2″ 14646 days ago?

In another theme that handled widgets, similar hanky, but with doing something with an array that PHP didn’t like.

Unfortunately, Bravenewcode, the authors, have their support forums in plastic surgery right now and I’m not sure how to contact them. I’ve also tried hitting them up on Twitter for a clue on how to hit them up. Nothing yet.

Update: the lads from Bravenewcode have quickly fixed this. It was an error caused by Twitter not returning an array when only one tweet was available (when it normally does). WordTwit 2.3.2 and WPTouch 1.9.8.1 address this issue.

Saving a local draft fails. Not iust fails, but completely doesn’t happen at all. Writing a post of any length goes into /dev/null. Seems it’s not an exactly uncommon issue as this forums post shows but for some reason isn’t reproduceable.

Saving as a draft to the server is fine, though.

I did a bit of digging around the site’s help pages and I’m sure it’s no surprise to anyone that it’s pretty much impossible to contact an actual person to help you out.  I do have to credit the CEO Mark Zuckerberg’s enormous ego with creating what is possibly one of the most successful sites with the paradoxically worst support system ever.  The customer forum is composed of an unbelievably byzantine interface link structure that rivals Microsoft’s.  It’s pretty much impossible to find the answer to a problem which they don’t deem to be an official “issue”, nevermind one they do.

Anyway, I did eventually find a solution (without input, of course, from any Facebook staff, possibly from Google).  For some reason, they’ve started scanning their database for contact emails that they deem “generic” (e.g. “mail@…” and “info@…”).  At their request, I changed mine to something else, something equally generic but not what they’re flagging and then set it on my email server side to forward to the address I had previously given Facebook.  This seemed to satisfy it.

I did notice in the “help” pages that some customers had tried to follow Facebook’s instructions but on doing so, became locked out of their accounts.  They’re now stuck in purgatory.  I won’t dwell on this too long, since it’s only a matter of time until I too, am banished with all my contacts, groups and fan pages blown to dust.

Seriously, Facebook:  F.O.

Also, anecdotally (from the “help” pages), more than one customer had reported finding a help page in there somewhere that directed them to contact Facebook support at the following address: “info@facebook.com”.

This had been happening for months and I started to get suspicious that it might be BT doing something sneaky like sending a reboot command to the router.  After all, they update the firmware remotely and silently with out telling you.  Who knows what else they can do?

I renewed my search for an answer a few days ago and came across this posting in a forum in which it’s reported that a British Telecom support person says that it’s a fault in the firmware (version 8.1.H.G) and should be fixed by the end of June (which it wasn’t).  There’s also this posting in which it’s suggested that it was a security feature programmed into the firmware.  To do what, exactly?

If it was, in fact, a security “feature” is crippling the router the best way to handle it?  Who exactly is being protected by the unit indiscriminately rebooting itself?  What’s the scenario in that programmer’s head that leads to the conclusion that killing my FTP transfers is the right thing to do?  Although, this smacks of middle management incompetence not having the scope to understand the ramifications of their actions when they say “do this dumb thing.”

The “feature” apparently only kicks in when you’re doing uploading with no download traffic (as if this is a tell-tale sign of perfidy) so just go download something while you’re uploading and maybe it will work.

When I restarted Firefox, no problem…weird.  I guess this sometimes happens?

I had to take the webapp.css file from the helvetireader.webapp (which is just a zip archive) and drop it into the Prism instance (at /Applications/Helvetireader/Contents/Resources/webapp/).  You could also just directly create this file – it’s contents are just a CSS import as described in that last link.

Unlike the Fluid-based SSB, Prism instances seem to require extensions, plugins and themes separately, i.e. per-instance.  I guess this is a good thing and avoids the kind of snarl that Gears and Fluid are having.

FluidInstance[3592]: Failed to create window context device
CGWindowContextCreate: failed to create context delegate.
_initWithWindowNumber: error creating graphics ctxt object for ctxt:0x18343, window:0xffffffff

The bug was being discussed here (in which commenters pointed their fingers at several plugins) but the discussion seems to have ceased.  The last suspected culprit is the Google Gears plugin (located in /Library/Internet Plug-Ins).  Indeed, when I removed every plugin and added them back, the Gears plugin was the one that caused the flooding.

Running a sample trace in Activity Monitor comes up with this:


16 WebNetscapePluginEventHandlerCarbon::sendNullEvent()
16 WebNetscapePluginEventHandlerCarbon::sendEvent(EventRecord*)
16 -[WebNetscapePluginDocumentView sendEvent:isDrawRect:]
16 -[WebNetscapePluginDocumentView saveAndSetNewPortStateForUpdate:]
14 -[NSWindow _threadContext]
14 -[NSWindowGraphicsContext _initWithWindowNumber:scaleFactor:]
7 CGWindowContextCreate
7 __CGWindowContextDelegateCreate
5 RIPDeviceCreateWindowDevice
5 CGSDeviceCreate
5 _CGSReferenceWindowByID
5 _CGSFindSharedWindow
5 _CGSGetSharedWindow
5 mach_msg
5 mach_msg_trap
5 mach_msg_trap

The Fluid developer doesn’t seem to have figured out if it’s a Fluid issue or a specific plugin issue.

Apparently, it’s a known problem and they’re trying to fix it, although their forums are somewhat unhelpful.  I’ve yet to see anything actually completely fixed, though.

Final Cut’s Compressor needs Qmaster to be running in the background to arbitrate rendering jobs – even if you’re the only machine on the network.  Lots of babies get thrown out with the bathwater when you do a operating system upgrade (including some X11 essentials like xinit and startx – more on that later), and migrating your big cat from Tiger to Leopard is no different.  Compressor is one of the casualties.  There are 4 files in /usr/sbin/ that get tossed: qmastera, qmasterd, qmasterprefs and qmasterqd.

If you did an Archive & Install, you can fish those babies out from /Previous Systems/<YYYY-MM-DD_XXXX>/usr/sbin, but you’ll probably need some Terminal skills.

Strangely enough, the startup launch scripts in /Library/StartupItems/Qmaster will probably still be there, so you can either reboot or again in the Terminal, “sudo /Library/StartupItems/Qmaster start” should get it running again.

Also, perhaps this works but I haven’t tried it.  Anyone?