Matthew Barnes [Sun, 15 Jul 2012 14:02:34 +0000 (10:02 -0400)]
online-accounts: Mail sources need to be writable.
Unfortunately there is not a clean separation between GOA-dictated
settings and user preferences in mail account ESource triplets, so they
all need to be writable and it's up to the client application to defend
the settings which must not be changed.
Matthew Barnes [Sun, 15 Jul 2012 14:01:18 +0000 (10:01 -0400)]
ECollectionBackend: Minor cleanup.
Ihar Hrachyshka [Sun, 15 Jul 2012 08:07:16 +0000 (11:07 +0300)]
Updated Belarusian translation.
Matthew Barnes [Sat, 14 Jul 2012 12:37:14 +0000 (08:37 -0400)]
Bug 559391 - Wrong number of unread mails in folder tree
This reverts part of the original fix for this bug which stopped reading
from the summary database and caused all kinds crazy message counts to
appear in Evolution's folder tree (often negative counts, or negative
counts cast to an unsigned integer).
Message counts are back to normal now.
Reverts part of commit
42489b26c45956c66186a5dc96083437cf7a579a.
Milan Crha [Fri, 13 Jul 2012 13:12:38 +0000 (15:12 +0200)]
Bug #578240 - Impossible to cancel search in search folders
Nilamdyuti Goswami [Fri, 13 Jul 2012 12:15:41 +0000 (17:45 +0530)]
Assamese translation updated
Chao-Hsiung Liao [Thu, 12 Jul 2012 01:04:21 +0000 (09:04 +0800)]
Updated Traditional Chinese translation(Hong Kong and Taiwan)
Milan Crha [Wed, 11 Jul 2012 10:29:49 +0000 (12:29 +0200)]
Preset timeout on SoupSession-s to 90 seconds
Milan Crha [Mon, 9 Jul 2012 14:50:03 +0000 (16:50 +0200)]
The default namespace is an empty string in IMAP settings
Piotr Drąg [Mon, 9 Jul 2012 14:11:50 +0000 (16:11 +0200)]
Updated POTFILES.in
Fran Diéguez [Mon, 9 Jul 2012 12:58:18 +0000 (14:58 +0200)]
Updated Galician translations
Milan Crha [Mon, 9 Jul 2012 09:57:02 +0000 (11:57 +0200)]
Bug #677886 - Runtime warning from ESourceAlarms
Matthew Barnes [Sun, 8 Jul 2012 19:56:22 +0000 (15:56 -0400)]
Camel: Use folder display name is status messages when possible.
And also put the folder name in quotes.
Matthew Barnes [Sun, 8 Jul 2012 19:41:22 +0000 (15:41 -0400)]
camel_folder_expunge_sync(): Push a status message.
Matthew Barnes [Sun, 8 Jul 2012 16:16:54 +0000 (12:16 -0400)]
CamelSubscribable: Connect before [un]subscribing.
Matthew Barnes [Sun, 8 Jul 2012 16:08:09 +0000 (12:08 -0400)]
CamelSubscribable: Push status messages for blocking operations.
Matthew Barnes [Sat, 7 Jul 2012 13:32:30 +0000 (09:32 -0400)]
Bug 678602 - Inform server of client-side auth cancellations.
Matthew Barnes [Sat, 7 Jul 2012 12:00:31 +0000 (08:00 -0400)]
server_side_source_remove(): Disregard G_IO_ERROR_NOT_FOUND.
Matthew Barnes [Sat, 7 Jul 2012 01:46:18 +0000 (21:46 -0400)]
CamelStoreClass: Rename compare_folder_name to equal_folder_name.
CamelStoreClass has the following method:
GCompareFunc compare_folder_name;
Both the method name and call signature imply this is a strcmp()-style
function where a zero return value means the values are equal. But in
fact this method is used as an equality test where a zero return value
means the values are NOT equal.
Rename the method and change its call signature like so:
GEqualFunc equal_folder_name;
This is an API break but clearly a necessary one.
Matthew Barnes [Fri, 6 Jul 2012 20:45:22 +0000 (16:45 -0400)]
evolution-source-viewer.c: Fix a frequent runtime warning.
Kjartan Maraas [Fri, 6 Jul 2012 17:08:44 +0000 (19:08 +0200)]
Updated Norwegian bokmål translation
Matthew Barnes [Fri, 6 Jul 2012 04:21:31 +0000 (00:21 -0400)]
Remove source_extension_constructed().
Was written before GObject implemented its own constructed() method.
Matthew Barnes [Fri, 6 Jul 2012 04:20:39 +0000 (00:20 -0400)]
CamelImapSettings: Change namespace default to empty string.
Matthew Barnes [Fri, 6 Jul 2012 04:13:31 +0000 (00:13 -0400)]
Revert "Bug #678901 - Fake ESource::changed notifications"
This reverts commit
c982babcd69d1059ecbe01d2de9bade4073d9859.
Breaks "changed" notifications in EServerSideSource and causes
Bug 678842 - Enable GMail GOA Calendar doesn't add it to UI
Matthew Barnes [Thu, 5 Jul 2012 12:55:34 +0000 (08:55 -0400)]
evolution-source-viewer: Fix icon column alignment.
Matthew Barnes [Thu, 5 Jul 2012 12:43:52 +0000 (08:43 -0400)]
evolution-source-viewer: Change "UID" label to "Identity".
So it's consistent with the tree view column.
Matthew Barnes [Thu, 5 Jul 2012 11:43:42 +0000 (07:43 -0400)]
evolution-source-viewer: Add delete capability.
Easy way to get rid of botched or unwanted test accounts.
Matthew Barnes [Wed, 4 Jul 2012 20:38:34 +0000 (16:38 -0400)]
Bug 678903 - GConf migration omits so-called "send-only" accounts
Matthew Barnes [Wed, 4 Jul 2012 16:25:32 +0000 (12:25 -0400)]
Disregard G_FILE_ERROR_NOENT during GConf migration.
Not finding GConf XML files is a good thing!
Milan Crha [Wed, 4 Jul 2012 09:51:20 +0000 (11:51 +0200)]
Bug #559391 - Wrong number of unread mails in folder tree
Alexander Shopov [Fri, 29 Jun 2012 05:17:24 +0000 (08:17 +0300)]
Updated Bulgarian translation
Debarshi Ray [Tue, 3 Jul 2012 11:39:43 +0000 (13:39 +0200)]
Fix make distcheck
Milan Crha [Tue, 3 Jul 2012 09:17:51 +0000 (11:17 +0200)]
Bug #678901 - Fake ESource::changed notifications
Dimitris Spingos [Mon, 2 Jul 2012 04:34:08 +0000 (07:34 +0300)]
Updated Greek translation
Matthew Barnes [Sun, 1 Jul 2012 12:50:10 +0000 (08:50 -0400)]
Remove redundancy from CRITICAL message.
Matthew Barnes [Sun, 1 Jul 2012 11:50:59 +0000 (07:50 -0400)]
camel_provider_list(): Emit CRITICAL if camel_provider_load() fails.
Otherwise CamelProvider modules may be silently dropped without so much
as a console message.
Nguyễn Thái Ngọc Duy [Sat, 30 Jun 2012 03:10:34 +0000 (10:10 +0700)]
Updated Vietnamese translation
Nguyễn Thái Ngọc Duy [Sat, 30 Jun 2012 02:57:18 +0000 (09:57 +0700)]
po/vi: import from Damned Lies
Matthew Barnes [Thu, 28 Jun 2012 15:08:18 +0000 (11:08 -0400)]
libedata-cal.pc.in: Add a 'libexecdir' variable.
Matthew Barnes [Thu, 28 Jun 2012 15:07:29 +0000 (11:07 -0400)]
libedata-book.pc.in: Add a 'libexecdir' variable.
Milan Crha [Wed, 27 Jun 2012 18:15:52 +0000 (20:15 +0200)]
Bug #677697 - Huge memory leaks
Giovanni Campagna [Sat, 24 Mar 2012 00:06:02 +0000 (01:06 +0100)]
weather: update for newer GWeather
GWeather 3.6 changed API to be properly namespaced and GObject based.
Also, WeatherLocation was removed and replaced with GWeatherLocation
in all public API (with additionally the possibility to retrieve one
particular GWeatherLocation by station code)
At the same time, remove a old and now useless abstration by folding
EWeatherSourceCCF into EWeatherSource.
https://bugzilla.gnome.org/show_bug.cgi?id=672805
Milan Crha [Wed, 27 Jun 2012 12:41:22 +0000 (14:41 +0200)]
Fix configure warning about BUILT_SOURCES
Matthew Barnes [Wed, 27 Jun 2012 12:27:54 +0000 (08:27 -0400)]
ECollectionBackend: Add default method for dup_resource_id().
Now checks for an E_SOURCE_EXTENSION_RESOURCE by default and returns the
"identity" value if the extension if present. Otherwise it returns NULL.
Matthew Barnes [Wed, 27 Jun 2012 12:18:15 +0000 (08:18 -0400)]
Add ESourceResource extension type.
Intended to be used by collection backends to note the server-assigned
name of a calendar, address book, or any other kind of remote resource
that an ESource can represent.
Matthew Barnes [Wed, 27 Jun 2012 12:15:52 +0000 (08:15 -0400)]
Avoid including <db.h> in a public header file.
Don't want Berkeley DB to be a requirement in libedata-book.pc.
Colin Walters [Tue, 26 Jun 2012 21:15:01 +0000 (17:15 -0400)]
build: libebook: Fix race condition in gen-western-table.py
In a parallel build, the normal compiler process will race with the
process writing to the table header. Depending on timing, it may or
may not work.
Fix this in two ways:
* Write the file atomically, so #include either gets the whole thing,
or nothing.
* Add it to BUILT_SOURCES so automake knows to build it first.
Daniel Mustieles [Tue, 26 Jun 2012 19:11:05 +0000 (21:11 +0200)]
Updated Spanish translation
Milan Crha [Tue, 26 Jun 2012 16:43:26 +0000 (18:43 +0200)]
Incorrect check for CamelSettings value change
This was causing runtime warnings on evolution start from IMAP provider,
which reads values from settings for junk/trash real folder, but
the settings object was set as NULL, which means "create default settings
for this class", but the test I added for object pointer change did
succeed for the NULL object (both internal and passed in settings are NULL
at the beginning), thus the default settings object was not created.
Christian Hilberg [Tue, 26 Jun 2012 14:35:57 +0000 (16:35 +0200)]
CamelIMAPXServer: use g_return_if_fail() family of functions instead of g_assert()
* changed the function param sanity checking from assertions
to the g_return*_if_fail() family of functions, so we can
stumble along instead of barfing in front of the user
Christian Hilberg [Tue, 26 Jun 2012 14:32:53 +0000 (16:32 +0200)]
IMAPX: (utils) use g_return_if_fail() family of functions instead of g_assert()
* changed the function param sanity checking from assertions
to the g_return*_if_fail() family of functions, so we can
stumble along instead of barfing in front of the user
Christian Hilberg [Fri, 22 Jun 2012 15:21:36 +0000 (17:21 +0200)]
CamelIMAPXServer: exposed command run function in public API
* in order to actually make use of the untagged response
handler extensions, CamelIMAPXStore derivatives will
need to send custom commands to the IMAP server
* the newly exposed API function provides a means for this
Christian Hilberg [Fri, 22 Jun 2012 15:16:50 +0000 (17:16 +0200)]
IMAPX: (utils) made capabilities flags extensible
* added a hash table as an IMAP capabilities flags
lookup mechanism for the capabilities parser util
function
* added a capabilities strings registration function
(to add new capabilities flags, which will then be
known to the parser and assigned a new flag bit.
This is returned by the registration function so
the caller can test whether the bit is set in the
CamelIMAPXServer capabilities bitstring)
Christian Hilberg [Thu, 14 Jun 2012 17:11:22 +0000 (19:11 +0200)]
CamelIMAPXServer: added documentation for untagged handler descriptor struct
Christian Hilberg [Thu, 14 Jun 2012 16:47:41 +0000 (18:47 +0200)]
CamelIMAPXServer: added untagged response handler registration API function
* register new untagged response handlers with this function
* can also be used to replace existing handlers
(think twice before doing that), while keeping track
of the original handler function
Christian Hilberg [Thu, 14 Jun 2012 14:56:46 +0000 (16:56 +0200)]
CamelIMAPXServer: do not hide possible programming issue
* do not free the CamelIMAPXServer private context
structure at the beginning of imapx_untagged()
as it is expected to be NULL
* if it isn't, the function got called concurrently
for the same CamelIMAPXServer instance, in which
case we need special locking mechanisms anyway
* thanks to Milan Crha for pointing this out
Christian Hilberg [Thu, 14 Jun 2012 14:21:06 +0000 (16:21 +0200)]
CamelIMAPXServer: use function table for untagged handler dispatching
* added a handler GHashTable to CamelIMAPXServer private data
* CamelIMAPXServer instance init populates the table with
an initial set of handler functions (the ones previously
existing in the class)
* using that function table in imapx_untagged() now to
find the handler function registered for a given untagged
response
Christian Hilberg [Thu, 14 Jun 2012 14:19:28 +0000 (16:19 +0200)]
IMAPX: added string definitions for untagged responses
Christian Hilberg [Wed, 13 Jun 2012 11:07:36 +0000 (13:07 +0200)]
CamelIMAPXServer: moved handler code for untagged responses to individual functions
* this is the first step towards an extensible CamelIMAPXServer
w.r.t. untagged IMAP responses
* there is now one handler function per untagged response
* instead of one big switch statement, there will be a
table with pointers to the new handler function, which
will be indexed by the untagged response code
* this will eventually eliminate the switch statement
in imapx_untagged()
* the gperf dependency will be dropped in the process
Matthew Barnes [Tue, 26 Jun 2012 14:36:34 +0000 (10:36 -0400)]
Fix install failure in google backend.
Instead of linking to libebook-google-utils.la, just add its source
files directly to the libebookbackendgoogle module. So the code is
duplicated in the backend module and the noinst unit test library.
This was causing libtool to barf on bogus undefined reference errors
during install. Still don't really understand what libtool's problem
was -- maybe it didn't like the noinst library? Who knows.
Milan Crha [Tue, 26 Jun 2012 12:02:40 +0000 (14:02 +0200)]
Bug #678856 - Crash in add_instance() at e-cal-client.c:1311
Nilamdyuti Goswami [Tue, 26 Jun 2012 07:58:31 +0000 (13:28 +0530)]
Assamese translation reviewed
Philip Withnall [Mon, 25 Jun 2012 21:34:35 +0000 (22:34 +0100)]
google: Remove an outdated workaround for an EBookBackendCache bug
The bug itself was fixed in commit
d535d4b09c61a8277084ddce0cfafc737b69fab2,
two years ago.
Philip Withnall [Mon, 25 Jun 2012 21:25:20 +0000 (22:25 +0100)]
google: Remove the option to not cache Google Contacts data locally
We always download the entire contacts list, so there’s little point in not
caching it on disk. This saves bandwidth and speeds up initial access to the
address book. The cache file shouldn’t get unreasonably big, given that users
will typically only have a few hundred contacts in their address book.
Philip Withnall [Mon, 25 Jun 2012 20:43:35 +0000 (21:43 +0100)]
vala: Add custom overrides for e_contact_[get|get_const|set]()
They were using void* in Vala, which is generally a bad idea. In this case,
it was resulting in unavoidable memory leaks in libfolks (bug #678781).
This changes the functions to be simple generics instead.
Matthew Barnes [Mon, 25 Jun 2012 01:45:52 +0000 (21:45 -0400)]
Post-release version bump.
Matthew Barnes [Mon, 25 Jun 2012 01:32:57 +0000 (21:32 -0400)]
NEWS update for 3.5.3 release.
Matthew Barnes [Sun, 24 Jun 2012 21:28:06 +0000 (17:28 -0400)]
Camel string pool improvements.
Keep the pointer to the allocated string and the allocated string's
reference count together in a new StringPoolNode struct. This way a
StringPoolNode can be used as both the key and value in a GHashTable,
allowing the table to save some memory. Also use the slab allocator
for the fixed-size StringPoolNode instances.
Didn't do any real formal measurements for this so the actual reduction
in memory usage may be insignificant. If nothing else the code is more
readable now, and manipulating reference counts is less awkward.
Fran Diéguez [Sun, 24 Jun 2012 21:24:43 +0000 (23:24 +0200)]
Updated Galician translations
Matthew Barnes [Sun, 24 Jun 2012 16:52:48 +0000 (12:52 -0400)]
Coding style and whitespace cleanup.
Matthew Barnes [Sun, 24 Jun 2012 12:18:46 +0000 (08:18 -0400)]
libebook: Reference address books by UID instead of URI.
Breaks EContact and EVCard APIs, but merely making the names accurate.
Philip Withnall [Sun, 24 Jun 2012 11:35:23 +0000 (12:35 +0100)]
google: Remove use of ESourceExtensionSecurity
Google have required the use of SSL connections to Google Contacts for a
while, and libgdata has used SSL connections exclusively for even longer.
The ‘use_ssl’ option was completely ineffective and redundant.
Philip Withnall [Sun, 24 Jun 2012 11:09:29 +0000 (12:09 +0100)]
google: Remove an unused variable
It was always the case that: live_mode == (priv->bookviews != NULL)
Philip Withnall [Sun, 24 Jun 2012 08:55:46 +0000 (09:55 +0100)]
google: Add a new custom vCard property containing Google system groups
This can then be picked up by libfolks (and other interested clients) to
allow them to detect which system groups a Google contact is in. Previously,
they had to infer this from the categories the contact was in, which are
localised. This was tricky and fragile.
Helps: https://bugzilla.gnome.org/show_bug.cgi?id=676383
Matthew Barnes [Sun, 24 Jun 2012 03:32:25 +0000 (23:32 -0400)]
Deprecate EDbHash.
Apparently nothing uses it anymore.
Matthew Barnes [Sun, 24 Jun 2012 00:26:10 +0000 (20:26 -0400)]
ESource: Use a collation key for display name comparisons.
e_source_compare_by_display_name() gets called a lot. We've been
calling g_utf8_collate() each time, which is known to be slow when
comparing large numbers of UTF-8 strings.
We now stash a collation key during e_source_set_display_name()
by passing the UTF-8 display name to g_utf8_collate_key(). Then
in e_source_compare_by_display_name() we can simply call strcmp()
on the internal collation keys to compare display names.
Didn't take any real performance measurements other than seeing
e_source_compare_by_display_name() topping the charts in sysprof.
Matthew Barnes [Sat, 23 Jun 2012 15:35:11 +0000 (11:35 -0400)]
Add missing linker flags for libebook-google-utils.
Sasi Bhushan Boddepalli [Sat, 23 Jun 2012 04:41:20 +0000 (10:11 +0530)]
Updated Telugu Translation
Matthew Barnes [Fri, 22 Jun 2012 21:00:29 +0000 (17:00 -0400)]
Bug 678067 - Localized display names get stuck
Matthew Barnes [Fri, 22 Jun 2012 20:10:06 +0000 (16:10 -0400)]
ESource: Minor code cleanup.
Matthew Barnes [Fri, 22 Jun 2012 13:43:30 +0000 (09:43 -0400)]
camel-provider.h: Fix a typo.
Matthew Barnes [Fri, 22 Jun 2012 13:25:24 +0000 (09:25 -0400)]
CAMEL_PROVIDER_IS_STORE_AND_TRANSPORT: Handle NULL gracefully.
Evaluates to FALSE if the 'provider' pointer is NULL.
Matthew Barnes [Fri, 22 Jun 2012 12:49:48 +0000 (08:49 -0400)]
ESourceContacts: Only include local address books by default.
A more reasonable default than including all address books by default.
Weeds out large, remote address books that can take forever to query.
Matthew Barnes [Fri, 22 Jun 2012 12:19:58 +0000 (08:19 -0400)]
Bug 678618 - Runtime warning when clicking on top-level sidebar item
Matthew Barnes [Thu, 21 Jun 2012 17:21:56 +0000 (13:21 -0400)]
source_idle_changed_cb: Avoid a runtime warning.
Apparently it's possible for the 'changed' GSource to already be NULL
when we go to unref it in source_idle_changed_cb(). Not sure how that's
happening... possibly racing with dispose()? In any case just check for
NULL first.
Matthew Barnes [Thu, 21 Jun 2012 12:54:29 +0000 (08:54 -0400)]
EServerSideSource: Don't write changes to disk until exported.
If a collection backend is configuring an EServerSideSource from a
worker thread, avoid writing changes to disk prematurely.
Explicitly calling e_source_write_sync() still works as advertised
whether the source is exported or not.
Matthew Barnes [Thu, 21 Jun 2012 12:40:54 +0000 (08:40 -0400)]
EServerSideSource: Add an "exported" property.
Read-only boolean property indicates whether the source has been
exported over D-Bus. Basically just queries the registry server
for itself.
Daniel Mustieles [Thu, 21 Jun 2012 10:29:30 +0000 (12:29 +0200)]
Updated Spanish translation
Matthew Barnes [Wed, 20 Jun 2012 19:43:02 +0000 (15:43 -0400)]
Fix build failures when $(top_srcdir) != $(top_builddir).
Really need to get distcheck working again...
Nilamdyuti Goswami [Wed, 20 Jun 2012 16:05:27 +0000 (21:35 +0530)]
Assamese translation reviewed
Milan Crha [Wed, 20 Jun 2012 08:16:25 +0000 (10:16 +0200)]
ESource: Read/Write int64/uint64
Matthew Barnes [Tue, 19 Jun 2012 21:14:05 +0000 (17:14 -0400)]
Improve imapx_command_fetch_message_done().
Using a corrupted database table as an opportunity to improve code
readability and error handling.
Matthew Barnes [Tue, 19 Jun 2012 21:03:07 +0000 (17:03 -0400)]
camel_data_cache_get: Improve empty file handling.
When an empty cache file is detected, set a GError with a descriptive
error message in addition to returning NULL.
Matthew Barnes [Tue, 19 Jun 2012 19:17:04 +0000 (15:17 -0400)]
e_book_backend_file_open: Fix directory existence test.
In e_book_backend_file_open() we seem to be bailing out if the data
directory does not exist, but yet we don't create the directory until
further down.
Not sure if this is correct, but it seems to me we only want to check
for the data directory if 'only_if_exists == TRUE'.
Matthew Barnes [Tue, 19 Jun 2012 18:36:46 +0000 (14:36 -0400)]
EDataBookFactory: Improve error message for GOA connection failure.
Fallout from a debugging session.
The printed error message was simply
libedata-book-WARNING **: The given address is empty
which was not very helpful in tracking down the source.
Matthew Barnes [Tue, 19 Jun 2012 15:44:13 +0000 (11:44 -0400)]
camel_data_cache_get_filename: Remove unused GError parameter.
Don't want this parameter to result in unnecessary error handling.
Matthew Barnes [Tue, 19 Jun 2012 14:44:31 +0000 (10:44 -0400)]
imapx_command_fetch_message_done: Fix an error handling issue.
Matthew Barnes [Tue, 19 Jun 2012 12:40:46 +0000 (08:40 -0400)]
Bug 678378 - ESourceRegistry causes deadlock in gnome-shell
Avoid scheduling idle callbacks while populating the registry for the
first time during initialization, and therefore having to iterate the
main context to force those idle callbacks to be dispatched.
Instead call source_registry_add_source() directly, just during the
initial population. This will emit "source-added" signals from the
object manager thread, but no one will be listening.
Matthew Barnes [Tue, 19 Jun 2012 12:22:58 +0000 (08:22 -0400)]
ESourceRegistry: Split ESource creation into a standalone function.
Split ESource creation out of the "object-added" signal handler to
a standalone function -- source_registry_new_source() -- which the
signal handler now calls.
Matthew Barnes [Tue, 19 Jun 2012 11:33:40 +0000 (07:33 -0400)]
ESourceRegistry: Use g_main_context_ref_thread_default().
Slightly more convenient than g_main_context_get_thread_default(), which
requires checking for NULL before referencing the main context ourselves.