Re: liferea: diff for NMU version 1.8.6-1+nmu1
On 11/10/2012 05:10 PM, Paul Gevers wrote:
On 10-11-12 09:46, David Smith wrote:
Why did you update libtool-dont-rearange-as-needed patch without
any functional change?
While inspecting / testing if it was still needed, I may have
inadvertently updated the date on the file. No changes in the file,
I have restored the original file in preparation for my next
attempt.
The following is noise:
---- liferea-1.8.3.orig/ltmain.sh
-+++ liferea-1.8.3/ltmain.sh
+--- a/ltmain.sh
++++ b/ltmain.sh
Thanks, removed.
Some of the other important bugs are very old (years) and I wasn't
able to reproduce them.
Please tag them as unreproducible [1]
Done
Some of them, I'm waiting for people to get back to me on if they
still impact the latest version of liferea.
I think you have tagged them with moreinfo, right?
Yes. Done.
The rest probably should be forwarded to upstream as I don't think
it's an easy fix if they're still valid bugs.
So you have added this information as well. Don't hesitate to forward
them (and tag them as such of course), but it would be nice to reproduce
if possible.
Done.
Are really all these bugs of the proper severity? I.e. bug 692007
does not seem to qualify in my view (although *you* marked it as
important later, but when you submitted you did not think it
important). You can try, but I am unsure if this would qualify for
an unblock (maybe in addition to the rest).
In my opinion, they are. Originally I was using this software with
only a few feeds. When I added more feeds, I realized this was a
more serious bug as it then started spamming the user's desktop with
notifications on feeds that have already been read. Further, they
keep triggering every refresh so it's pretty much endless. I
suppose the user could just turn the notifications off for the app,
but I thought it better to just fix it. Liferea is a news reader
and people use it to get live updates to newstreams via RSS so I
personally wouldn't use the package without the patch so I thought
it should be "important".
I understand. It is just that the release-team might disagree. But
off-course you can try. Next time, (or even now as a comment) add this
kind of justification to the command where you raise the severity.
OK. I've gone back and now added justifications for the bugs that I made
important and included the fixes for in this NMU.
Some of your patches change unnecessary things such as whitespace,
capitalization or copyright dates. If I was release manager I
would prefer you leave the changes to the functional parts. You can
add the copyright notice with the date to the header of the patch
instead, if you want. I think added comments to added code are
fine.
I pretty much took those slightly bloated (with comments/whitespace)
patches directly from upstream. I wasn't sure if it was ok to simply
not include parts of upstream patches. I could trim them down a bit
and move the copyright to the patch header if that's preferred or
makes it easier to see what has changed.
The release team made it clear that changes should be clean, i.e. your
changes of getting an unblock are higher if you remove everything that
is not necessary.
* Cleaned up the names of the patches so they more clearly describe what
they accomplish.
* Moved the upstream changes of the copyright to the patch header so the
patch is a lot smaller.
* I also added a patch for adding the hardened build wrapper. Liferea
has a parser that parses feeds from the Internet so I thought it best to
harden the compile, and this also fixes two lintian warnings. I was
told on IRC that hardened packages is a release goal so they should
automatically be "important" , but I was also told that using the
hardened wrapper to do it is deprecated. It was suggested that using the
other way (build flags) to harden the package may be too invasive as the
package is only compat 7 and has it's own custom build flags.. So I'm
not sure what to do here if simply adding the hardened build wrapper
isn't acceptable.
Going to try my best to get everything through to Wheezy, but if it
doesn't I suppose I can go back and just remove what didn't make it and
try again.
Attached is the new NMU diff.
Also note that I uploaded to mentors, filled out the template and sent
it in. I am trying to get a sponsor for this NMU that is mostly crash
fixes from upstream.
Sponsorship request and latest changelog is here:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=692975
Target is Wheezy.
diff -Nru liferea-1.8.6/debian/changelog liferea-1.8.6/debian/changelog
--- liferea-1.8.6/debian/changelog 2012-07-01 05:45:40.000000000 +0800
+++ liferea-1.8.6/debian/changelog 2012-11-11 22:07:08.000000000 +0800
@@ -1,3 +1,22 @@
+liferea (1.8.6-1.1) unstable; urgency=low
+
+ * Non-maintainer upload.
+ * Fixed crash when dragging Google Reader feeds to parent node.
+ (Closes: #692526)
+ * Fixed crash when removing folders in Google Reader feeds. (Closes: #692525)
+ * Fixed crash when opening empty links in feeds. (Closes: #692272)
+ * Fixed crash when network online status changes. (Closes: #692270)
+ * Fixed bug where web browser doesn't launch or the wrong web browser is
+ launched. (Closes: #668197)
+ * Fixed bug where user's desktop was spammed with notifications about
+ Google Reader feeds that have already been read. (Closes: #692007)
+ * Replaced build-depends on transitional package libwebkit-dev with
+ libwebkitgtk-dev. (Closes: #677749)
+ * Added hardening-wrapper since liferea has a parser and should be
+ built with hardening. (Closes: #692527)
+
+ -- David Michael Smith <sidicas2@gmail.com> Sat, 10 Nov 2012 10:43:14 +0800
+
liferea (1.8.6-1) unstable; urgency=low
* Fixes ever growing temporary DB files. (patch by Sven Hartge)
diff -Nru liferea-1.8.6/debian/control liferea-1.8.6/debian/control
--- liferea-1.8.6/debian/control 2012-07-01 05:38:09.000000000 +0800
+++ liferea-1.8.6/debian/control 2012-11-11 22:03:10.000000000 +0800
@@ -15,14 +15,15 @@
libglade2-dev,
libgtk2.0-dev (>= 2.18.0),
libglib2.0-dev (>= 2.24.0),
- libwebkit-dev (>= 1.2.2),
+ libwebkitgtk-dev (>= 1.8.1),
intltool,
libicu-dev,
libx11-dev,
dh-autoreconf,
libice-dev,
libunique-dev,
- libjson-glib-dev
+ libjson-glib-dev,
+ hardening-wrapper
Standards-Version: 3.9.3
Homepage: http://liferea.sourceforge.net/
Vcs-Browser: http://git.debian.org/?p=collab-maint/liferea.git
diff -Nru liferea-1.8.6/debian/patches/fix-browser-selections liferea-1.8.6/debian/patches/fix-browser-selections
--- liferea-1.8.6/debian/patches/fix-browser-selections 1970-01-01 08:00:00.000000000 +0800
+++ liferea-1.8.6/debian/patches/fix-browser-selections 2012-11-10 20:04:47.000000000 +0800
@@ -0,0 +1,35 @@
+Patch by David Smith <sidicas2@gmail.com>
+Fixed a bug where web browser doesn't launch due to not
+having gnome desktop installed, or the wrong web browser is
+launched due to not having an appropriate fallback.
+Posted by him to #668197
+Last-Update 2012-11-07
+
+--- a/src/browser.c
++++ b/src/browser.c
+@@ -137,7 +137,7 @@
+ }
+
+ if (fallback && !cmd) /* Last fallback: first browser default */
+- cmd = browsers[0].defaultplace;
++ cmd = browsers[6].defaultplace;
+ }
+ g_free (libname);
+
+@@ -244,6 +244,7 @@
+ struct browser *browser;
+ gchar *cmd = NULL;
+ gboolean done = FALSE;
++ gchar *libname;
+
+ g_assert (uri != NULL);
+
+@@ -256,6 +257,8 @@
+ g_free (cmd);
+ } else {
+ /* the "default" browser has no command to use the GTK launch mechanism */
++ conf_get_str_value (BROWSER_ID, &libname);
++ if (g_str_equal (libname, "default"))
+ done = gtk_show_uri (NULL, uri, 0, NULL);
+ }
+ }
diff -Nru liferea-1.8.6/debian/patches/fix-crash-in-empty-launch-url liferea-1.8.6/debian/patches/fix-crash-in-empty-launch-url
--- liferea-1.8.6/debian/patches/fix-crash-in-empty-launch-url 1970-01-01 08:00:00.000000000 +0800
+++ liferea-1.8.6/debian/patches/fix-crash-in-empty-launch-url 2012-11-10 20:04:47.000000000 +0800
@@ -0,0 +1,36 @@
+From 31fced473a8351749405fae99872556059b308ae Mon Sep 17 00:00:00 2001
+From: Matthias Maennich <matthias@maennich.net>
+Date: Wed, 15 Aug 2012 00:06:29 +0200
+Subject: [PATCH] fix crash in liferea_webkit_launch_url(...) for empty links
+ in feeds
+
+in case of empty links in feeds liferea crashed with the
+following backtrace:
+ #0 in __strstr_sse42 () from /lib/libc.so.6
+ #1 in liferea_webkit_launch_url (scrollpane=<>, url=0x0) at webkit.c:515
+ #2 in itemlist_selection_changed (item=item@entry=<>) at itemlist.c:599
+
+this patch fixes the crash with showing the default itemview content
+instead of trying to launch the browser with the non-existing URL
+Reported-by: Matthias Maennich <matthias@maennich.net>
+
+--- a/src/itemlist.c
++++ b/src/itemlist.c
+@@ -599,8 +599,15 @@
+ if (node->loadItemLink) {
+ gchar* link = item_make_link (item);
+
+- itemview_launch_URL (link, TRUE /* force internal */);
+- g_free (link);
++ // in case of an empty link entry the above results in link==NULL
++ // so, only show the link if exists
++ if (link) {
++ itemview_launch_URL (link, TRUE /* force internal */);
++ g_free (link);
++ } else { // else show the normal itemview content
++ itemview_select_item(item);
++ itemview_update();
++ }
+ } else {
+ itemview_select_item (item);
+ itemview_update ();
diff -Nru liferea-1.8.6/debian/patches/fix-crash-when-dragging-feeds-outside-google-reader liferea-1.8.6/debian/patches/fix-crash-when-dragging-feeds-outside-google-reader
--- liferea-1.8.6/debian/patches/fix-crash-when-dragging-feeds-outside-google-reader 1970-01-01 08:00:00.000000000 +0800
+++ liferea-1.8.6/debian/patches/fix-crash-when-dragging-feeds-outside-google-reader 2012-11-10 20:04:47.000000000 +0800
@@ -0,0 +1,21 @@
+From 5f245fe87e8f32ff0e13c2e94addaf6032b806c4 Mon Sep 17 00:00:00 2001
+From: Lars Windolf <lars.lindner@gmail.com>
+Date: Mon, 30 Jul 2012 23:26:01 +0200
+Subject: [PATCH] Fixes SF #2855990: Crash when dragging Google Reader feeds outside Google Reader. This is now prevented. (reported by algnod)
+
+--- a/src/ui/ui_dnd.c
++++ b/src/ui/ui_dnd.c
+@@ -116,6 +116,13 @@
+ if (gtk_tree_model_get_iter (GTK_TREE_MODEL (model), &iter, src_path)) {
+ gtk_tree_model_get (GTK_TREE_MODEL (model), &iter, FS_PTR, &sourceNode, -1);
+
++ g_assert (sourceNode);
++
++ /* Never drop into another node source as this arises to many problems
++ (e.g. remote sync, different subscription type, e.g. SF #2855990) */
++ if (NODE_SOURCE_TYPE (targetNode) != NODE_SOURCE_TYPE (sourceNode))
++ return FALSE;
++
+ if (IS_FOLDER(sourceNode) && !(NODE_SOURCE_TYPE (targetNode)->capabilities & NODE_SOURCE_CAPABILITY_HIERARCHIC_FEEDLIST))
+ return FALSE;
+ }
diff -Nru liferea-1.8.6/debian/patches/fix-crash-when-online-status-changes liferea-1.8.6/debian/patches/fix-crash-when-online-status-changes
--- liferea-1.8.6/debian/patches/fix-crash-when-online-status-changes 1970-01-01 08:00:00.000000000 +0800
+++ liferea-1.8.6/debian/patches/fix-crash-when-online-status-changes 2012-11-10 20:04:47.000000000 +0800
@@ -0,0 +1,25 @@
+From 41d21ec9d844ddf62b8e71b0476ab4cffca32654 Mon Sep 17 00:00:00 2001
+From: Yanko Kaneti <yaneti@declera.com>
+Date: Fri, 14 Sep 2012 16:13:00 +0300
+Subject: [PATCH] Fix crash when network online status changes
+
+From Fedora bug https://bugzilla.redhat.com/show_bug.cgi?id=857348 :
+"liferea frequently crashes when manipulating htmlviews (opening/closing
+new tabs) while the network online status changes. This is caused
+because the signal notifying online status is not properly disconnected
+when the htmlview is removed."
+---
+ src/ui/liferea_htmlview.c | 2 ++
+ 1 file changed, 2 insertions(+)
+
+--- a/src/ui/liferea_htmlview.c
++++ b/src/ui/liferea_htmlview.c
+@@ -77,6 +77,8 @@
+ static void
+ liferea_htmlview_finalize (GObject *object)
+ {
++ g_signal_handlers_disconnect_by_data (network_monitor_get (), object);
++
+ G_OBJECT_CLASS (parent_class)->finalize (object);
+ }
+
diff -Nru liferea-1.8.6/debian/patches/fix-crash-when-removing-google-reader-folders liferea-1.8.6/debian/patches/fix-crash-when-removing-google-reader-folders
--- liferea-1.8.6/debian/patches/fix-crash-when-removing-google-reader-folders 1970-01-01 08:00:00.000000000 +0800
+++ liferea-1.8.6/debian/patches/fix-crash-when-removing-google-reader-folders 2012-11-11 00:16:07.000000000 +0800
@@ -0,0 +1,157 @@
+Debian Bug# 692525. Fixes crash when removing folders in Google Reader feeds.
+
+From 739a294f7c045c37e9dbce97455a71cb2357e068 Mon Sep 17 00:00:00 2001
+From: Lars Windolf <lars.lindner@gmail.com>
+Date: Fri, 5 Oct 2012 12:09:42 +0200
+Subject: [PATCH] Fixes SF #3572660: crash in google_source_remove_node (reported by Yanko Kaneti)
+
+--- a/src/fl_sources/default_source.c
++++ b/src/fl_sources/default_source.c
+@@ -210,6 +210,8 @@
+ .description = "The default feed list source. Should never be added manually. If you see this then something went wrong!",
+ .capabilities = NODE_SOURCE_CAPABILITY_IS_ROOT |
+ NODE_SOURCE_CAPABILITY_HIERARCHIC_FEEDLIST |
++ NODE_SOURCE_CAPABILITY_ADD_FEED |
++ NODE_SOURCE_CAPABILITY_ADD_FOLDER |
+ NODE_SOURCE_CAPABILITY_WRITABLE_FEEDLIST,
+ .source_type_init = default_source_init,
+ .source_type_deinit = default_source_deinit,
+--- a/src/fl_sources/node_source.h
++++ b/src/fl_sources/node_source.h
+@@ -50,11 +50,12 @@
+
+ enum {
+ NODE_SOURCE_CAPABILITY_IS_ROOT = (1<<0), /**< flag only for default feed list source */
+- NODE_SOURCE_CAPABILITY_MULTI_INSTANCES = (1<<1), /**< allows multiple source instances */
+- NODE_SOURCE_CAPABILITY_DYNAMIC_CREATION = (1<<2), /**< feed list source is user created */
+- NODE_SOURCE_CAPABILITY_WRITABLE_FEEDLIST = (1<<3), /**< the feed list tree of the source can be changed */
+- NODE_SOURCE_CAPABILITY_HIERARCHIC_FEEDLIST = (1<<4), /**< the feed list tree of the source can have hierarchic folders */
+- NODE_SOURCE_CAPABILITY_ITEM_STATE_SYNC = (1<<5) /**< the item state can and should be sync'ed with remote */
++ NODE_SOURCE_CAPABILITY_DYNAMIC_CREATION = (1<<1), /**< feed list source is user created */
++ NODE_SOURCE_CAPABILITY_WRITABLE_FEEDLIST = (1<<2), /**< the feed list tree of the source can be changed */
++ NODE_SOURCE_CAPABILITY_ADD_FEED = (1<<3), /**< feeds can be added to the source */
++ NODE_SOURCE_CAPABILITY_ADD_FOLDER = (1<<4), /**< folders can be added to the source */
++ NODE_SOURCE_CAPABILITY_HIERARCHIC_FEEDLIST = (1<<5), /**< the feed list tree of the source can have hierarchic folders */
++ NODE_SOURCE_CAPABILITY_ITEM_STATE_SYNC = (1<<6) /**< the item state can and should be sync'ed with remote */
+ };
+
+ /** feed list node source type */
+--- a/src/node.c
++++ b/src/node.c
+@@ -494,6 +494,28 @@
+ return baseUrl;
+ }
+
++gboolean
++node_can_add_child_feed (nodePtr node)
++{
++ g_assert (node->source->root);
++
++ if (!(NODE_TYPE (node->source->root)->capabilities & NODE_CAPABILITY_ADD_CHILDS))
++ return FALSE;
++
++ return (NODE_SOURCE_TYPE (node)->capabilities & NODE_SOURCE_CAPABILITY_ADD_FEED);
++}
++
++gboolean
++node_can_add_child_folder (nodePtr node)
++{
++ g_assert (node->source->root);
++
++ if (!(NODE_TYPE (node->source->root)->capabilities & NODE_CAPABILITY_ADD_CHILDS))
++ return FALSE;
++
++ return (NODE_SOURCE_TYPE (node)->capabilities & NODE_SOURCE_CAPABILITY_ADD_FOLDER);
++}
++
+ /* node children iterating interface */
+
+ void
+--- a/src/node.h
++++ b/src/node.h
+@@ -355,6 +355,24 @@
+ */
+ const gchar * node_get_base_url(nodePtr node);
+
++/**
++ * Query wether a feed be added to the given node.
++ *
++ * @param node the node
++ *
++ * @returns TRUE if a feed can be added
++ */
++gboolean node_can_add_child_feed (nodePtr node);
++
++/**
++ * Query wether a folder be added to the given node.
++ *
++ * @param node the node
++ *
++ * @returns TRUE if a folder can be added
++ */
++gboolean node_can_add_child_folder (nodePtr node);
++
+ /* child nodes iterating interface */
+
+ typedef void (*nodeActionFunc) (nodePtr node);
+--- a/src/ui/popup_menu.c
++++ b/src/ui/popup_menu.c
+@@ -262,14 +262,15 @@
+ ui_popup_node_menu (nodePtr node, gboolean validSelection, guint button, guint32 activate_time)
+ {
+ GtkWidget *menu;
+- gboolean writeableFeedlist, isRoot, isHierarchic;
++ gboolean writeableFeedlist, isRoot, isHierarchic, addChildren;
+
+ menu = gtk_menu_new ();
+
+ if (node->parent) {
+ writeableFeedlist = NODE_SOURCE_TYPE (node->parent->source->root)->capabilities & NODE_SOURCE_CAPABILITY_WRITABLE_FEEDLIST;
+- isRoot = NODE_SOURCE_TYPE (node->parent->source->root)->capabilities & NODE_SOURCE_CAPABILITY_IS_ROOT;
++ isRoot = NODE_SOURCE_TYPE (node->source->root)->capabilities & NODE_SOURCE_CAPABILITY_IS_ROOT;
+ isHierarchic = NODE_SOURCE_TYPE (node->parent->source->root)->capabilities & NODE_SOURCE_CAPABILITY_HIERARCHIC_FEEDLIST;
++ addChildren = NODE_TYPE (node->source->root)->capabilities & NODE_CAPABILITY_ADD_CHILDS;
+ } else {
+ /* if we have no parent then we have the root node... */
+ writeableFeedlist = TRUE;
+@@ -287,7 +288,7 @@
+ }
+
+ if (writeableFeedlist) {
+- if (NODE_TYPE (node->source->root)->capabilities & NODE_CAPABILITY_ADD_CHILDS) {
++ if (addChildren) {
+ GtkWidget *item;
+ GtkWidget *submenu;
+
+@@ -295,9 +296,10 @@
+
+ item = ui_popup_add_menuitem (menu, _("_New"), NULL, NULL, NULL, 0);
+
+- ui_popup_add_menuitem (submenu, _("New _Subscription..."), ui_popup_add_feed, NULL, NULL, 0);
++ if (node_can_add_child_feed (node))
++ ui_popup_add_menuitem (submenu, _("New _Subscription..."), ui_popup_add_feed, NULL, NULL, 0);
+
+- if (isHierarchic)
++ if (node_can_add_child_folder (node))
+ ui_popup_add_menuitem (submenu, _("New _Folder..."), ui_popup_add_folder, NULL, NULL, 0);
+
+ if (isRoot) {
+@@ -309,7 +311,7 @@
+ gtk_menu_item_set_submenu (GTK_MENU_ITEM(item), submenu);
+ }
+
+- if (node->children) {
++ if (isRoot && node->children) {
+ gtk_menu_shell_append (GTK_MENU_SHELL(menu), gtk_separator_menu_item_new());
+ ui_popup_add_menuitem (menu, _("Sort Feeds"), ui_popup_sort_feeds, node, GTK_STOCK_SORT_ASCENDING, 0);
+ }
+--- a/src/fl_sources/google_source.c
++++ b/src/fl_sources/google_source.c
+@@ -360,6 +360,7 @@
+ "present your Google Reader subscriptions, and will synchronize your feed list and reading lists."),
+ .capabilities = NODE_SOURCE_CAPABILITY_DYNAMIC_CREATION |
+ NODE_SOURCE_CAPABILITY_WRITABLE_FEEDLIST |
++ NODE_SOURCE_CAPABILITY_ADD_FEED |
+ NODE_SOURCE_CAPABILITY_ITEM_STATE_SYNC,
+ .source_type_init = google_source_init,
+ .source_type_deinit = google_source_deinit,
diff -Nru liferea-1.8.6/debian/patches/fix-google-reader-notifications liferea-1.8.6/debian/patches/fix-google-reader-notifications
--- liferea-1.8.6/debian/patches/fix-google-reader-notifications 1970-01-01 08:00:00.000000000 +0800
+++ liferea-1.8.6/debian/patches/fix-google-reader-notifications 2012-11-10 20:04:48.000000000 +0800
@@ -0,0 +1,18 @@
+Patch by David Smith <sidicas2@gmail.com>
+Prevent Google Reader feeds from receiving "unread" status from
+google reader if the local status is "read". This fixes a bug
+where user's desktop was spammed with notifications about
+google reader feeds that have already been read.
+Posted by him to: #692007
+Last-Update: 2012-11-05
+
+--- a/src/itemset.c
++++ b/src/itemset.c
+@@ -168,6 +168,7 @@
+ which means online accounts we sync against, but not normal
+ online feeds where items have no read status. */
+ if (allowStateChanges) {
++ if ((!oldItem->readStatus) && (newItem->readStatus))
+ oldItem->readStatus = newItem->readStatus;
+ oldItem->flagStatus = newItem->flagStatus;
+ }
diff -Nru liferea-1.8.6/debian/patches/series liferea-1.8.6/debian/patches/series
--- liferea-1.8.6/debian/patches/series 2012-02-27 09:54:47.000000000 +0800
+++ liferea-1.8.6/debian/patches/series 2012-11-10 20:04:48.000000000 +0800
@@ -1,3 +1,9 @@
www-browser
debian-example-feeds
libtool-dont-rearange-as-needed
+fix-browser-selections
+fix-crash-in-empty-launch-url
+fix-crash-when-dragging-feeds-outside-google-reader
+fix-crash-when-online-status-changes
+fix-crash-when-removing-google-reader-folders
+fix-google-reader-notifications
diff -Nru liferea-1.8.6/debian/rules liferea-1.8.6/debian/rules
--- liferea-1.8.6/debian/rules 2012-02-27 09:54:47.000000000 +0800
+++ liferea-1.8.6/debian/rules 2012-11-10 21:04:04.000000000 +0800
@@ -5,6 +5,7 @@
CFLAGS += -g -O$(if $(findstring noopt,$(DEB_BUILD_OPTIONS)),0,2)
LDFLAGS += "-Wl,--as-needed"
export LIBS=-lX11 -lICE
+export DEB_BUILD_HARDENING=1
# These are used for cross-compiling and for saving the configure script
# from having to guess our platform (since we know it already)
Reply to: