Liferea

Logged In: YES
user_id=754825

That did it indeed.
BTW it's kind of nice to update the .spec file :-) IMHO the
first few lines of it should be :
Summary: Liferea (Linux RSS News Aggregator)
Name: liferea
Version: 0.4.7d
Release: 2

Logged In: YES
user_id=952951

I looks like on some entries, the 2nd 0.4.7d package has
problems displaying the content on some feeds. I'll attach
the cache file for a feed that has this problem.
Specifically, it has problems with the 4/29 entry.

Logged In: YES
user_id=834800

I fixed the problem with the missing headlines.
I replace the 0.4.7d tarball. Please try the new
one.

Logged In: YES
user_id=952951

In addition to problem that George is getting, I'm still
getting Segfaults when updating all feeds. When I upgraded
to 0.4.7d, I removed all prior feeds, and readded them.

Here's a snippet from gdb:

downloading "http://www.gnome.org/~seth/blog/?flav=rss"

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1084447520 (LWP 19874)]
0x00c1b04c in _int_free () from /lib/tls/libc.so.6

Logged In: YES
user_id=754825

Yes, indeed the 0.4.7d doesn't crash anymore, but now it
gets no hedadlines (just the date displayed in article list
pane amd Item empty) for (among others)
http://slashdot.org/index.rss.

Logged In: YES
user_id=834800

Should be fixed in 0.4.7d.

Logged In: YES
user_id=834800

I finally found time to have a look at the
cache file. Nathans fix should work well.
Each time the parser reads invalid contents
that cannot be recovered it returns NULL.
As you can see in dir2_19 the first articles
have correctly encoded contents, the rest
has no entities. This is due to a failed loading
of this XML file in the recovery mode of the
parser. libxml2 does only extract the ASCII
characters if the tags are totally broken.

The reason for this might be not properly
encoded characters retrieved from the source
feed which were not fixed during the UTF-8
conversions.

With Nathans fix the cache loading/saving itself
is correct. If such a bug appears again it would
be very helpful to use strace to determine the
cache file which caused the crash and to post
the current version of the source feed and the
cache file here! So we could debug the converting.

The fix will be included in the 0.4.7d I plan to
release tonight. Thanks for your patience!

Logged In: YES
user_id=416550

There was a problem in the cache file. I have no idea how it
got there. I made a quick patch so that Liferea does not
crash, but the issue should be investigated further. The
HTML encoding of the cache file is very wrong. I have
attached a patch that will fix the issue for now.

Logged In: YES
user_id=754825

I cannot attach a file (it seems) here. So I've sent it via
e-mail. I have had a reply from Lars Lindner saying he
didn't forgot me :-)

Logged In: YES
user_id=416550

(I'm not sure if Lars Lindner has responded to your emails
in the last week and fixed the bug.) Could you either post
the file "/home/kgeorge/.liferea/dir2_19" on this bug or
send it to me? My email address is t98502 at
users.sourceforge.net.

Logged In: YES
user_id=754825

Unfortunately I seem to not be able to attach files to the
bug. So sent it by email to llando at users.sourceforge.net.

Here are the last ~20 lines :

open("/etc/ld.so.cache", O_RDONLY) = 17
open("/usr/lib/libpng12.so.0", O_RDONLY) = 17
open("/usr/share/liferea/pixmaps/flag.png", O_RDONLY) = 16
open("/usr/share/liferea/pixmaps/available.png", O_RDONLY) = 16
open("/usr/share/liferea/pixmaps/unavailable.png", O_RDONLY)
= 16
open("/usr/share/liferea/pixmaps/ocs.png", O_RDONLY) = 16
open("/usr/share/liferea/pixmaps/directory.png", O_RDONLY) = 16
open("/usr/share/liferea/pixmaps/help.png", O_RDONLY) = 16
open("/usr/share/liferea/pixmaps/vfolder.png", O_RDONLY) = 16
open("/usr/lib/gconv/ISO8859-1.so", O_RDONLY) = 17
open("/usr/share/liferea/pixmaps/empty.png", O_RDONLY) = 16
open("/home/kgeorge/.liferea/dir2_21", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_20", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_4", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_29", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_29.xpm", O_RDONLY) = 16
open("/home/kgeorge/.liferea/dir2_16", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_28", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_1", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_2", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_3", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_3.xpm", O_RDONLY) = 16
open("/home/kgeorge/.liferea/dir2_5", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_5.xpm", O_RDONLY) = 16
open("/home/kgeorge/.liferea/dir2_6", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_7", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_8", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_8.xpm", O_RDONLY) = 16
open("/home/kgeorge/.liferea/dir2_9", O_RDONLY|O_LARGEFILE) = 16
open("/home/kgeorge/.liferea/dir2_10", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_11", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_11.xpm", O_RDONLY) = 16
open("/home/kgeorge/.liferea/dir2_12", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_13", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_14", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_15", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_15.xpm", O_RDONLY) = 16
open("/home/kgeorge/.liferea/dir2_17", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_18", O_RDONLY|O_LARGEFILE)
= 16
open("/home/kgeorge/.liferea/dir2_19", O_RDONLY|O_LARGEFILE)
= 16
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
+++ killed by SIGSEGV +++

Logged In: YES
user_id=834800

Thanks for the info that the original problem
seems to be fixed! I tried to reproduce the
gaim feed problem but was not successful.
The only oddity is see that a subscription
to this feed causes mktime() warning.

If you don't mind could you zip you ~/.liferea
cache directory and attach it to the bug? Then
I could check the cache files. It would also be
interesting to see which was the last opened
file. So maybe you could run once more with

strace -e trace=open /usr/local/bin/liferea-bin

and post the last 20 lines or so?

If you want to fix the problem to continue using
the program please move (or delete) the cache
directory ~/.liferea, start the program and remove
the gaim feed subscription (if it is the cause of
the problem!). After that you could move the
cache back and everything should work fine.

Logged In: YES
user_id=754825

I just added the following RSS :
http://gaim.sourceforge.net/rss.php/news
and it now crashes each time I start the liferea :
$ liferea
Mozilla is not available...
available browser modules (/usr/lib/liferea):
-> GtkHTML2 (liblihtmlg.so)
No browser module configured!
trying to load browser module GtkHTML2 (liblihtmlg.so)
Use of deprecated SAXv1 function endElement
Segmentation fault (core dumped)

Stacktrace in gdb :
(gdb) where
#0 0x0052797a in __strtol_internal () from /lib/tls/libc.so.6
#1 0x08067dd5 in parseCacheItem ()
#2 0x08059fa3 in saveAllFeeds ()
#3 0x0805a069 in addFeed ()
#4 0x0805737a in loadSubscriptions ()
#5 0x08067f1a in main ()

Logged In: YES
user_id=754825

Recompiled & running the 0.4.7c-test2. The Liferea works
allmost half a day now and no crashes, no hangs.
It seems resolved. Thanks.

Logged In: YES
user_id=834800

0.4.7b contains some RSS parsing bugs.
These will be fixed in 0.4.7c which you
can already try using the test tarball
available from the project homepage.

http://liferea.sf.net/liferea-0.4.7c-test2.tar.gz

0.4.7c will come within this week. If one of
you could test against it that would be great :-)

Logged In: NO

Hi.

I see a very similar instability on my computer using
Liferea 0.4.7b. My system is a SuSE 9.0, with KDE upgraded
to the 3.2.1 release.

Liferea has got big problems when I've got a Directory
subscription. I tried with a few of them, and whenever it
tries to update them it 's got a segmentation fault.

Very difficult to get rid of it when I let it check
subscriptions on startup... ;-) If I remove the directories
I get a good stability again.

Thanks for the attention.

Marco
marco.montagnana@prometeia.it

Logged In: YES
user_id=754825

I also get a GPF with 0.4.7b (something I wasn't getting
with the 0.4.7-test5 and the 0.4.6. Here's my stack trace
(Fedora Core 1):
(gdb) where
#0 0x0072b01b in strlen () from /lib/tls/libc.so.6
#1 0x006f9ba5 in vfprintf () from /lib/tls/libc.so.6
#2 0x0071a8d6 in vasprintf () from /lib/tls/libc.so.6
#3 0x003b28b0 in g_strdup_vprintf () from
/usr/lib/libglib-2.0.so.0
#4 0x003b2926 in g_strdup_printf () from
/usr/lib/libglib-2.0.so.0
#5 0x08055c93 in bufferParseError ()
#6 0x001d19e1 in xmlEncodeEntitiesReentrant () from
/usr/lib/libxml2.so.2
#7 0x001f832c in xmlNsListDumpOutput () from
/usr/lib/libxml2.so.2
#8 0x001f7cf2 in xmlNsListDumpOutput () from
/usr/lib/libxml2.so.2
#9 0x001f7f3c in xmlNsListDumpOutput () from
/usr/lib/libxml2.so.2
#10 0x001f7cf2 in xmlNsListDumpOutput () from
/usr/lib/libxml2.so.2
#11 0x001f7f3c in xmlNsListDumpOutput () from
/usr/lib/libxml2.so.2
#12 0x001f7cf2 in xmlNsListDumpOutput () from
/usr/lib/libxml2.so.2
#13 0x001f7f3c in xmlNsListDumpOutput () from
/usr/lib/libxml2.so.2
#14 0x001f8687 in xmlNodeDumpOutput () from
/usr/lib/libxml2.so.2
#15 0x001f9b1e in xmlSaveFormatFileEnc () from
/usr/lib/libxml2.so.2
#16 0x08059ac8 in saveFeed ()
#17 0x08054a5a in checkForUpdateResults ()
#18 0x0806e778 in argset_free ()
#19 0x003a28ec in g_main_context_wakeup () from
/usr/lib/libglib-2.0.so.0
#20 0x0039ffdd in unblock_source () from
/usr/lib/libglib-2.0.so.0
#21 0x003a0fd8 in g_main_context_dispatch () from
/usr/lib/libglib-2.0.so.0
#22 0x003a12ef in g_main_context_dispatch () from
/usr/lib/libglib-2.0.so.0
#23 0x003a19cf in g_main_loop_run () from
/usr/lib/libglib-2.0.so.0
#24 0x0053744f in gtk_main () from /usr/lib/libgtk-x11-2.0.so.0
#25 0x0906dcb0 in ?? ()
#26 0xbfe8ed34 in ?? ()
#27 0xbfe8ed3c in ?? ()
#28 0xbfe8eca8 in ?? ()
#29 0x0906dcb0 in ?? ()

Logged In: YES
user_id=952951

Here's another trace from gdb:

** (liferea-bin:25552): WARNING **: Invalid UTF8 string
passed to pango_layout_set_text()
"Warren Togami" has no new items.
"Scott Collins" has not changed since last update.
"Online Help Feed" has not changed since last update.
"Fedora Linux / Fedora Core 1" has no new items.
"Christian Schaller" has no new items.
"Havoc Pennington" has not changed since last update.

** (liferea-bin:25552): WARNING **: Invalid UTF8 string
passed to pango_layout_set_text()
"OSNews" has 8 new items.

** (liferea-bin:25552): WARNING **: Invalid UTF8 string
passed to pango_layout_set_text()
"Robert Love" has not changed since last update.
"Slashdot" has no new items.

** (liferea-bin:25552): WARNING **: Invalid UTF8 string
passed to pango_layout_set_text()
"#tvtorrents" has not changed since last update.
"Wander" has not changed since last update.
"Joe Shaw" has no new items.
"Defensive Belligerence" has not changed since last update.
"T. Vander Stichele" has no new items.

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread -1084410656 (LWP 25552)]
0x0805b13d in freeFeed (fp=0x9255c28) at feed.c:924
924 ((struct feed_request
*)fp->request)->fp = NULL;

Logged In: YES
user_id=952951

I can reproduce this everytime I use the Update All Feeds.
I removed the Fedora feed, and it still segfaults.

Logged In: YES
user_id=834800

Some questions: is this 100% reproducable
or does it happen only sometimes? And does
it also happen when you update the Fedora
feed multiple times?