In 1.3b2, in addition to getting lots of failures on importing images and PDFs, I’m finding that it’s taking it a VERY long time to fail on an image (an image which looks fine under the Finder, btw). It can take it a full minute to finally fail on an image that has no problems. Any idea what’s going on? Anything I can do to help debug?
John
Ok, I have a simple test case which shows this problem in both 1.2 and 1.3b2:
Note: In my preferences, I have everything checked as “to be imported” except for Applescripts.
John
John, I didn’t time it, but it took less than ten seconds to index the 43 images into my database. (I’d say about 10 seconds elapsed between the time I hit the File > Index command and then looked for the new group.)
All of the captures were successful; I looked at five or so of the images, and the Log contained no error reports.
I’m running DTPO 1.3beta2 on a MacBook Pro 2.0 GHz with 2 GB RAM, under OS X 10.4.8.
You might want to run some maintenance operations on your OS and disk directory.
I’ve been having image import problems for about 2 weeks now. In fact, I found a PDF which causes my DTP to hang and never return:
OK, I really need to get this working again. I can believe that it’s not a problem in DTP, but how can I debug it? I’ll try doing full disk maintenance. Perhaps I have some kind of addon in the system that’s colliding. Do you have clue what might be causing a hang-up like this?
Note that if I turn off image importing, I have almost no problems at all – save for the PDF given above.
John
p.s. Thanks again for the quick response.
It now appears that attempting to Index any PDF causes DEVONthink to go into an infinite loop (no CPU time). I gave it about 30 minutes and it wouldn’t finish, on a simple PDF that I know it’s indexed before.
I’ll start doing my best to analyze it here.
John
John, that PDF that you can’t capture was captured into my database essentially instantly.
You might think about any software you have downloaded within the past few weeks. Christian commented in a similar case that third-party QuickTime plugins could cause problems.
Try removing such items to see if the problems go away, then perhaps (if you need them) add them back in one at a time to see if you can identify the trouble-maker.
I’ve disabling the following so far, to no effect:
/Library/
Internet Plug-Ins
LaunchDaemons
PDF Services
PreferencePanes
QuickTime
StartupItems
I don’t use personal QuickTime addons.
I’ll just keep narrowing it down. I also have another PowerBook, so hopefully I can do a system comparison that will lead some answers.
I also repaired both volumes on my machine. One thing I did find: that during the enormous pause before failing, there is no CPU activity at all.
John
A few other notes:
I am now unable to import any PDF or image file. Also, I’ve proved that I can open these same images in QuickTime instantly; they also import into iPhoto instantly.
I have the following sample data, taken for 20 seconds during the 28 seconds it takes for an image import to fail (and this amount of time is consistent).
John
Analysis of sampling pid 303 every 10.000000 milliseconds
Call graph:
2000 Thread_0f07
2000 0x274d
2000 0x2826
2000 NSApplicationMain
2000 -[NSApplication run]
2000 -[NSApplication nextEventMatchingMask:untilDate:inMode:dequeue:]
2000 _DPSNextEvent
2000 BlockUntilNextEventMatchingListInMode
2000 ReceiveNextEventCommon
2000 RunCurrentEventLoopInMode
2000 CFRunLoopRunInMode
2000 CFRunLoopRunSpecific
2000 __NSFireDelayedPerform
2000 0x1d259
2000 0x1158c4
2000 0x114514
1992 0x1186fa
1992 _objc_msgForward
1992 -[NSObject(NSForwardInvocation) forward::]
1992 -[NSConnection sendInvocation:]
1992 CFRunLoopRunInMode
1992 CFRunLoopRunSpecific
1992 mach_msg_trap
1992 mach_msg_trap
8 0x118645
8 -[HelperTask launch]
5 +[NSThread(NSThreadDevonExtensions) sleepTimeInterval:]
5 usleep
5 mach_wait_until
5 mach_wait_until
2 _objc_msgForward
2 -[NSObject(NSForwardInvocation) forward::]
2 -[NSConnection sendInvocation:]
2 CFRunLoopRunInMode
2 CFRunLoopRunSpecific
2 mach_msg_trap
2 mach_msg_trap
1 -[NSConcreteTask launch]
1 fork
1 fork
2000 Thread_1003
2000 _pthread_body
2000 __ape_agent
2000 mach_msg_trap
2000 mach_msg_trap
2000 Thread_1103
2000 _pthread_body
2000 forkThreadForFunction
2000 +[NSURLCache _diskCacheSyncLoop:]
2000 CFRunLoopRunInMode
2000 CFRunLoopRunSpecific
2000 mach_msg_trap
2000 mach_msg_trap
2000 Thread_1203
2000 _pthread_body
2000 CMMConvTask(void*)
2000 pthreadSemaphoreWait(t_pthreadSemaphore*)
2000 semaphore_wait_signal_trap
2000 semaphore_wait_signal_trap
1322 Thread_1303
1322 _pthread_body
1322 wait4
1322 wait4
678 Thread_1403
678 _pthread_body
678 wait4
678 wait4
Total number in stack (recursive counted multiple, when >=5):
5 _pthread_body
Sort by top of stack, same collapsed (when >= 5):
mach_msg_trap 5994
semaphore_wait_signal_trap 2000
wait4 2000
mach_wait_until 5
Oh, and I perhaps failed to mention: I tried back-versioning my system down to DTP 1.2, but the same problem exists. So it’s not something new in 1.3b2, but some kind of evil interaction between newly installed additions to my own system. 
John
Guess which directory I had to disable in order for everything to work:
~/Library/Caches/ShapeShifter
Because apparently having ShapeShifter UI mods active (which get deactivated if this directory is disabled) is causing a bad conflict with DTP, and the importing or indexing of images and PDFs.
What’s strange is that this problem did not appear as soon as I installed ShapeShifter. In fact, it worked for a while, but only gradually started to fail, until today things failed completely. Also, I know other people who use DT with ShapeShifter, and have no problems (yet).
Since I don’t have the tools to figure out exactly why these two are conflicting, ShapeShifter is going out the window.
Thanks for trying out the images I posted, Bill, you’re great.
John
Over time ShapeShifter has been involved in a variety of problems, and we really wish people would not install it, as it does nothing useful.
I’m cautious about things that deliberately modify Apple’s carefully designed operating system. Modifications involve potential stability risks; I don’t think modifying appearance justifies the risk, especially if one needs to get work done on the computer.