Seg Fault

[geormes]

Using copy paste run method 0.8.59 on Fedora gives
/usr/bin/xdg-icon-resource: line 589: 3207 Segmentation fault $GTK_UPDATE_ICON_CACHE -f -t "$1" > /dev/null 2> /dev/null


Previous versions worked great.

[lbolton]

Yup. me too.

Opensuse12.1/64

[kovidgoyal]

That appears to be a segfault in xdg-icon-resource, not calibre. So I am confused. Post the full log from the terminal.

[geormes]

Originally Posted by kovidgoyal

Post the full log from the terminal.

Wish I could. I removed directory and restarted and now the install "appears" to complete, but Calibre fails to start and freezes with initial screen covering 1/3 of workspace. Same as b4.

[geormes]

Originally Posted by geormes

Wish I could. I removed directory and restarted and now the install "appears" to complete, but Calibre fails to start and freezes with initial screen covering 1/3 of workspace. Same as b4.

Forgot to mention that I have gone back to .58

[lbolton]

Took a while to get back to this...

Opensuse 12.1,
KDE 4.7.2, Composition is OFF.
Calibre 0.8.60 (happened under 0.8.59 also)

Starting with v0.8.58 installed in /opt.
A normal install using the download script will continue through
the desktop integration section, and then throw the exception

/usr/bin/xdg-icon-resource: line 589: 3207 Segmentation fault $GTK_UPDATE_ICON_CACHE -f -t "$1" > /dev/null 2> /dev/null

At this point, ~home/.config/calibre/global.py has been overwritten.

Further attempts have the same result.

If global.py is edited to set database path and library path,
an Install will complete.

However, on first run, caliber still presents a dialog asking for a library path.
When the fields are filled, and the dialog closed,
the Dialog box seems to hang and the console gets:

GLib-CRITICAL **: g_main_context_pop_thread_default: assertion `g_queue_peek_head (stack) == context' failed

This is likely because the new installation still has a new uuid.
it is strange that having a partially updated global.py, seems to let calibre
avoid a broken code branch long enough to put up the dialog.

'Twould seem that, at this point, the uuid is updated.
Because any subsequent runs show the same behavior.
UNLESS, I again edit in those 2 paths.

All subsequent runs using:

calibre-debug --gui

will dump to console, messages much like the following;


--------------------------------------------cut---------------------------------------
shire:~ # calibre-debug --gui
Starting up...
Started up in 6.74 seconds with 10029 books
Job: 1 Get device information started
DeviceJob: 1 Get device information done, calling callback
DeviceJob: 1 Get device information callback returned
Job: 2 Get list of books on device started
DEBUG: 0.0 USBMS: Fetching list of books from device. Device= FOLDER_DEVICE oncard= None
DEBUG: 0.0 USBMS: dirs are: /local/staging/
DEBUG: 0.0 USBMS: scan from root False /local/staging/
DEBUG: 0.0 USBMS: count found in cache: 43, count of files in metadata: 43, need_sync: False
DEBUG: 0.0 USBMS: Finished fetching list of books from device. oncard= None
DEBUG: 0.1 USBMS: Fetching list of books from device. Device= FOLDER_DEVICE oncard= carda
DEBUG: 0.1 USBMS: Fetching list of books from device. Device= FOLDER_DEVICE oncard= cardb
DeviceJob: 2 Get list of books on device done, calling callback
Job: 1 Get device information finished
No details available.
Job: 2 Get list of books on device finished
No details available.
DeviceJob: 2 Get list of books on device callback returned
Job: 2 Get list of books on device finished
No details available.
Worker Launch took: 0.23487496376
InputFormatPlugin: EPUB Input running
on /local/books/James Patrick Kelly/10 16 to 1 (4098)/10 16 to 1 - James Patrick Kelly.epub
Found HTML cover titlepage.xhtml
QPainter::begin: Paint device returned engine == 0, type: 2
QPainter::setCompositionMode: Painter not active
QPainter::begin: Paint device returned engine == 0, type: 2
QPainter::begin: Paint device returned engine == 0, type: 2
QPainter::setCompositionMode: Painter not active
QPainter::begin: Paint device returned engine == 0, type: 2
/usr/bin/xdg-open: line 322: test: KDE Development Platform: 4.7.2 (4.7.2) "release 5": integer expression expected
/usr/bin/xdg-open: line 323: test: KDE Development Platform: 4.7.2 (4.7.2) "release 5": integer expression expected
/usr/bin/xdg-open: line 324: test: KDE Development Platform: 4.7.2 (4.7.2) "release 5": integer expression expected

(process:11870): GLib-CRITICAL **: g_main_context_pop_thread_default: assertion `g_queue_peek_head (stack) == context' failed
Worker Launch took: 0.245404005051
Job: 4 Convert book 1 of 1 (10 16 to 1) finished
Convert book 1 of 1 (10 16 to 1)
Resolved conversion options
calibre version: 0.8.60
{'asciiize': False,


.................deleted----------------


No TOC, MOBI index not generated
MOBI output written to /tmp/calibre_0.8.60_tmp_bbWOxU/QZ_OuQ.mobi


(process:11848): GLib-CRITICAL **: g_main_context_pop_thread_default: assertion `g_queue_peek_head (stack) == context' failed

(process:11848): GLib-CRITICAL **: g_main_context_pop_thread_default: assertion `g_queue_peek_head (stack) == context' failed
shire:~ #
-------------------------------------------------cut-------------------------------------

That's about as much debug I'm likely to get.


2. additional questions Please.

First, is seems a well known secret to everyone other than me, But,
How do I avoid having my Caliber environment overwritten every week.
It may have something to do with a python variable env_keep.
But, I'm having no luck understanding how to set and keep it.

Second, My attempt to install 2 versions side by side, and switch using a link in /opt/calibre/calibre, fails because python won't write to a link.
Is there a workaround for that?

TIA

[kovidgoyal]

You are likely suffering from the same libglib incompatibility described here: https://bugs.launchpad.net/bugs/1022019