Shane Synan [Sat, 3 Dec 2016 02:05:20 +0000 (20:05 -0600)]
Optionally use system locale for chat timestamp
Using QLocale::system().timeFormat(), check if the AM/PM designator
exists; if so, assume a 12-hour style timestamp, otherwise assume the
previous 24-hour style timestamp.
Generate a timestamp of either " hh:mm:ss" or " h:mm:ss ap" in order
to include seconds (QLocale::ShortFormat does not specify seconds).
Add new setting UseCustomTimestampFormat to switch between system
or user timestamp string. Bump settings version minor to keep old
behavior for upgrades.
Shane Synan [Fri, 2 Dec 2016 23:04:41 +0000 (17:04 -0600)]
Change ChannelMsg default to match system default
Change ChannelMsg default to match the system text color palette.
On systems tested, "palette(text)" doesn't refer to pure black, but
instead a very dark gray (60, 60, 60). Use that as default instead.
This won't affect existing installs as ChannelMsg is stored as
non-default if non-default colors are chosen.
Shane Synan [Sun, 4 Dec 2016 20:44:48 +0000 (14:44 -0600)]
Make ErrorMsg color distinct from ServerMsg
Switch ErrorMsg to a reddish color rather than the golden hues of
ServerMsg, matching the behavior of Quasseldroid.
This helps with discovering if Auto Identify fails now that SASL
authentication errors are reported in the server buffer. It also
points out any messages Quassel doesn't handle correctly, which might
be unexpected, but should help with discovering them.
No settings migration logic included as custom colors are an
all-or-nothing approach. Due to 'default.qss' using a system
palette color, there's no easy way to upgrade without changing
existing colors or loading all of Qt's color palettes in startup.
For example, palette(text) is dependent upon the system theme:
ChatLine {
foreground: palette(text);
}
Shane Synan [Fri, 2 Dec 2016 23:02:33 +0000 (17:02 -0600)]
Change CommandMsg color from purple to gray
The intense purple stands out a lot more than it probably needs to.
Let's switch that to a gray color.
No settings migration logic included as custom colors are an
all-or-nothing approach. Due to 'default.qss' using a system
palette color, there's no easy way to upgrade without changing
existing colors or loading all of Qt's color palettes in startup.
For example, palette(text) is dependent upon the system theme:
ChatLine {
foreground: palette(text);
}
Jonathan Riddell [Fri, 21 Oct 2016 16:07:52 +0000 (17:07 +0100)]
Switch from kdelibs4 sounds to Plasma 5 sound files
Resolves GH-260.
Marcus Eggenberger [Wed, 15 Mar 2017 21:38:07 +0000 (22:38 +0100)]
Merge pull request #277 from mamarley/psqlleast
Fix lastseenmsgid updates in PostgreSQL
Michael Marley [Thu, 9 Mar 2017 15:46:14 +0000 (10:46 -0500)]
Fix lastseenmsgid updates in PostgreSQL
GH-273 introduced a workaround for the issue where sometimes the
lastseenmsgid for a buffer was set to a msgid not in that buffer.
However, the workaround was incorrect as the "min()" function in
PostgreSQL is an aggregate function for use on a column, not a
function to return the least of a list of values passed in. The
correct function for use here is "least()".
Marcus Eggenberger [Mon, 6 Mar 2017 09:01:26 +0000 (10:01 +0100)]
Merge pull request #276 from digitalcircuit/fix-sql-postgres-newerthan
core: Fix Postgres messagesNewerThan wrong buffer
Shane Synan [Sun, 5 Mar 2017 19:39:38 +0000 (13:39 -0600)]
core: Fix Postgres messagesNewerThan wrong buffer
Make the inner SELECT statement use the buffer ID for bufferid, NOT
the messageid.
This fixes select_messagesNewerThan using the wrong variable for the
buffer ID, causing more recent messages to not show up when fetching
backlog with the "Unread messages" method.
Fixes regression in pull request #273, missed by pull request #274.
NOTE: The client still needs fixed to stop sending invalid IDs. That
should be addressed with another set of commits.
See https://github.com/quassel/quassel/pull/274#issuecomment-
284205802
Marcus Eggenberger [Sun, 5 Mar 2017 11:24:57 +0000 (12:24 +0100)]
Merge pull request #274 from digitalcircuit/fix-sql-inconsistencies
Fix SQLite setup, PostgreSQL migrate, cleanup SQL
Shane Synan [Sat, 4 Mar 2017 22:15:13 +0000 (16:15 -0600)]
Fix SQLite buffer setup, fix PostgreSQL migration
Remove the ALTER TABLE command, this isn't needed during setup as the
buffer's getting created for the first time.
Add the CHECK constraint as seen in the upgrade script, limiting the
value of lastseenmsgid.
Add handling for lastmsgid when migrating from SQLite to PostgreSQL.
Fix 'SELECT populate_lastmsgid()' call by first resetting the query.
This fixes breakages caused by incomplete work in pull request #273.
Shane Synan [Sat, 4 Mar 2017 21:09:14 +0000 (15:09 -0600)]
Add lastmsgid limit on fetching backlog to SQLite
Copy lastmsgid fetch limit improvements from PostgreSQL to SQLite,
improving performance by not requiring scanning from the start of the
backlog table to the present.
This builds upon pull request #273, including the original intent of
improving performance for unread backlog fetching. The original pull
request left out these lines when applying the PostgreSQL changes to
SQLite.
NOTE: The client-side generating these invalid values should be
fixed in future commits.
Shane Synan [Sun, 5 Mar 2017 04:58:00 +0000 (22:58 -0600)]
Unify SQL filenames, sort sql.qrc
Unify SQL filenames, renaming PostgreSQL 'select_messages' to
'select_messagesNewestK', and renaming SQLite 'select_messages' to
'select_messagesRange'.
Sort sql.qrc alphabetically.
This makes it easier to see which functionality is related, e.g. when
making future schema changes. Sorting improves 'git diff' output.
Shane Synan [Sun, 5 Mar 2017 04:57:51 +0000 (22:57 -0600)]
Add documentation, clean up whitespace
Maintain tabs in .sql files, remove extra newline.
Add documentation to the setLastMsg function.
This helps maintain consistency with the other files.
Marcus Eggenberger [Sat, 4 Mar 2017 14:37:15 +0000 (15:37 +0100)]
Merge pull request #273 from ReimuHakurei/master
Improve performance of PostgreSQL database in instances where buffers exist with no recent activity
Alex Ingram [Sat, 4 Mar 2017 02:34:20 +0000 (20:34 -0600)]
Port the previous PostgreSQL improvements to SQLite3.
Alex Ingram [Fri, 3 Mar 2017 20:05:30 +0000 (14:05 -0600)]
PostgreSQL: Improve performance and squish bug.
Improve performance of the unread messages backlog fetcher by using a normal JOIN instead of a LEFT JOIN.
Correct for bug in which buffer.lastseenmsgid was set to a value higher than buffer.lastmsgid. This is a client bug, but this prevents it from screwing up the database.
Alex Ingram [Fri, 3 Mar 2017 16:34:23 +0000 (10:34 -0600)]
Improve PostgreSQL performance for unread message fetch method, too.
Alex Ingram [Fri, 3 Mar 2017 05:40:23 +0000 (23:40 -0600)]
Improve performance of PostgreSQL for large databases.
This change requires a change to the database schema.
Currently, in cases where buffers exist that have not had any activity for an extended period of time, during initial backlog fetching, Quassel will have to scan backwards through all activity in all buffers until it reaches activity for the buffer in question.
This adds an additional field to the buffer table, "lastmsgid", which is the messageid of the most recent message added for that buffer. This improves performance by allowing Quassel to skip straight to where the most recent activity was, instead of scanning millions of rows (in large databases) in futility.
Shane Synan [Tue, 27 Dec 2016 13:26:03 +0000 (08:26 -0500)]
Split messages on newlines as per IRC protocol
When handling user input sent over the protocol, split apart messages
on QChar::LineFeed, processing and displaying each as individual
messages sent one after the other. This applies to /query, /say,
/notice, and /me. Otherwise, the IRC protocol interprets anything
after each line break as a new message.
Fixes inability to send multi-line content via copy-paste on
Quasseldroid. Quassel desktop currently works around this issue by
sending multiple individual messages for every line break.
Test case:
1. Install Quasseldroid (stable or alpha), connect to a core
2. Copy some text to the Android clipboard that has a line break in
it (not just word-wrap).
3. Paste into Quasseldroid
4. Send message
Before: Message shows up with line break (LF character on desktop),
IRC server treats all lines after first as raw commands.
After: Message gets split into multiple smaller messages at each
line break.
This should also work for /query, /say, /notice, and /me.
See: https://tools.ietf.org/html/rfc1459#page-8
Closes GH-263.
Willem Jan Palenstijn [Sun, 27 Nov 2016 13:34:07 +0000 (14:34 +0100)]
Make reloadCerts also handle the IPv6 server
Closes GH-264.
Hannah von Reth [Sat, 22 Oct 2016 13:37:55 +0000 (15:37 +0200)]
Do a full install by default
Closes GH-261.
Daniel Albers [Mon, 28 Nov 2016 18:20:45 +0000 (19:20 +0100)]
Fix AppVeyor Chocolatey failures by retrying
Merges pull request #265 from digitalcircuit/fix-ci-retry-chocolatey
Shane Synan [Mon, 28 Nov 2016 11:46:31 +0000 (05:46 -0600)]
Fix typo Chocolately -> Chocolatey
Official name is Chocolatey, e.g. https://chocolatey.org/ .
(The number of times I've mistyped this myself...)
Shane Synan [Mon, 28 Nov 2016 11:39:49 +0000 (05:39 -0600)]
Retry Chocolatey install to workaround failures
Automatically retry up to 3 times on failure when installing
Chocolatey modules, using the 'appveyor-retry' script. This can be
removed later if needed.
Sometimes Chocolatey returns a 404 for a package that exists. This
happens a lot. The official fix is to get the Business edition for
the more-stable private CDN, but since we're an open-source community
effort, sometimes a little kludgery goes a long way.
See http://help.appveyor.com/discussions/suggestions/816-generic-wrapper-for-retry#comment_40579488
And https://github.com/appveyor/ci/issues/418
Shane Synan [Thu, 13 Oct 2016 08:39:14 +0000 (03:39 -0500)]
Fix hang for single result search after others
Add check for equal start and end indices in updateHighlights, stop
the search loop when this happens. Regular search still seems to
work fine (if you found this via 'git blame', whoops).
Unfortunately, this seems specific to the contents of a buffer.
First, find a buffer that you've encountered freezing, and keep track
of what was loaded, where it was, and the two most recent search
terms.
For example...
1. Load some backlog to a buffer
2. Search for term with any number of matches
3. Making sure to -type over existing words without first
backspacing-, search for another term with only one match
Expected: Search results found, no freezing
Actual: Quassel hangs. startPos and endPos = same place,
start = 0, end = 0, _currentHighlight appears to retain
values from the previous search.
Resolves GH-259.
Shane Synan [Tue, 4 Oct 2016 06:54:42 +0000 (01:54 -0500)]
Re-add execute permissions to bundled inxi
Re-add execute permissions to bundled copy of inxi, letting Quassel
run it via /exec (including /sysinfo and /inxi).
Finishes up upgrade started in
83923d7c121dbcf514d4ff5b0ac2abb6eb08c6b9
Resolves GH-258.
Shane Synan [Fri, 30 Sep 2016 22:42:13 +0000 (17:42 -0500)]
Fix CoreNetwork::updateRateLimiting comments
Expand and fix opening summary comment, also remarking that the token
bucket should -not- be reset in this function.
Fix description of logic for _skipMessageRates.
Fix alignment of indented comments describing variables.
Resolves GH-257.
Shane Synan [Thu, 29 Sep 2016 00:28:28 +0000 (19:28 -0500)]
Hide AutoWHO for nicks without IrcUser objects
Don't require an IrcUser to exist when checking if WHO replies should
be hidden. The nickname is enough for checking.
This mimics CoreSessionEventProcessor::processIrcEvent403, where
ERR_NOSUCHCHANNEL replies are hidden if nickname is in the list
regardless of whether or not an IrcUser object exists.
Potentially fixes AutoWHO replies showing in the server buffer when
nicks quickly join and leave channels, resulting in the IrcUser
object getting destroyed before the server replies to the AutoWHO
request.
It's difficult to recreate the issue on demand, so it's quite
possible there's more to fix.
Resolves GH-256.
Martin T. H. Sandsmark [Sun, 25 Sep 2016 14:18:35 +0000 (16:18 +0200)]
Disable high DPI scaling on anything but Windows
The automatic scaling only works well on Windows.
See e. g.
e33532e048fa12ee32429ca83ee31aa8f065147d in qttools,
QTBUG-50698 and QTBUG-52318.
Resolves GH-254.
Shane Synan [Sat, 24 Sep 2016 06:59:05 +0000 (01:59 -0500)]
Enable per-chat history, line wrap by default
Enable input widget per-chat history and line wrapping by default,
following the behavior of Quassel Web and most Messaging/SMS apps
(HexChat does per-chat history by default, but not line wrapping).
Bump settings minor version to 3, preserve old defaults for upgrades.
This shows the minimum necessary to change default settings while
preserving previous defaults for existing installs.
Resolves GH-253.
Shane Synan [Thu, 22 Sep 2016 01:24:48 +0000 (20:24 -0500)]
Keep old clients from removing message rate delay
Add a guard against zero message delay in setMessageRateDelay,
following the error-checking of setMessageRateBurstSize. This blocks
old clients from setting message rate delay to 0 seconds for newly-
created networks, improperly overriding the default of 2.2 seconds.
This affects creating a network in an old client, switching to a new
client, then enabling custom rate-limits.
The supported method for removing rate limits by setting Unlimited to
true still works.
Resolves GH-252.
Ben Rosser [Mon, 19 Sep 2016 04:41:21 +0000 (00:41 -0400)]
Default to sqlite backend in core creation wizard
This introduces the notion of a "default" storage backend into
the core by adding a function (Core::isStorageBackendDefault),
which returns true if a given Storage object is 'default' and
false if not. This information is then added to the BackendInfo
dictionary passed through the protocol to the user interface,
which ensures the backend that is default will always be the
one displayed by default.
This change was prompted by conversation in IRC with [Saint]
and @digitalcircuit about making core creation slightly more
intuitive.
This allows us to, potentially, change the default storage
backend shown at core creation time should at some point
another backend be introduced, without having to modify
the interface.
It is a little weird having "isDefault" be an attribute of
*each* storage backend, but this was the simplest way to do
it without modifying the handshake. (Whichever is the last
storage backend to claim to be the default will win).
(As an aside, should this be merged and should #170 be merged
I'll implement something similar for authenticators in a
separate PR).
Resolves GH-250.
Manuel Nickschas [Sat, 8 Oct 2016 15:06:13 +0000 (17:06 +0200)]
dcc: Add settings page for DCC configuration
Manuel Nickschas [Fri, 7 Oct 2016 22:40:46 +0000 (00:40 +0200)]
dcc: Add persistent settings for core-side DCC
This introduces a new syncable object DccConfig that holds core-side
DCC settings, such as the network configuration. The configuration
is persisted as a per-user setting in the database.
The CoreSession's DccConfig instance is synced to connected clients
and accessible via the Client singleton.
Manuel Nickschas [Fri, 7 Oct 2016 22:30:03 +0000 (00:30 +0200)]
dcc: Clean up TransferManager a bit
Manuel Nickschas [Thu, 6 Oct 2016 17:42:38 +0000 (19:42 +0200)]
common: Work around SFINAE issues in MSVC
MSVC 2015 still has... issues with SFINAE, so the usual way to use
std::enable_if doesn't compile. So let's try and SFINAE this in a
different way until MS fixes their compiler.
Manuel Nickschas [Wed, 5 Oct 2016 17:29:19 +0000 (19:29 +0200)]
dcc: Name TransferManager object for syncing
This magically fixes older clients who try to sync an unnamed
TransferManager with unknown data types.
Manuel Nickschas [Tue, 4 Oct 2016 21:01:40 +0000 (23:01 +0200)]
common: Remove unused Exception type
Also fix a typo in docs.
Manuel Nickschas [Tue, 4 Oct 2016 20:53:01 +0000 (22:53 +0200)]
common: Add generic stream operators for enum types
These operators use the underlying type of a given enum to
(de)serialize to and from QDataStream.
Manuel Nickschas [Mon, 3 Oct 2016 23:25:18 +0000 (01:25 +0200)]
dcc: Retrieve transfer list on TransferManager change
If TransferModel is given a new manager, it should retrieve the
list of transfers just in case (even though, in practice, the manager
should not hold any transfers right after sync).
Manuel Nickschas [Thu, 29 Sep 2016 19:23:03 +0000 (21:23 +0200)]
dcc: Always clean up if a transfer completed or failed
Manuel Nickschas [Wed, 28 Sep 2016 21:04:17 +0000 (23:04 +0200)]
dcc: Properly sync TransferManager
Existing transfers were not synced to the client. With this commit,
the list of transfer IDs is marked as a Q_PROPERTY and thus synced
automatically.
Also provides the necessary serialization support.
Manuel Nickschas [Wed, 28 Sep 2016 21:03:27 +0000 (23:03 +0200)]
dcc: Only show file receive dialog for new transfers
... and not for ones that have been around on the core already
prior to syncing.
Manuel Nickschas [Wed, 28 Sep 2016 21:02:50 +0000 (23:02 +0200)]
dcc: Protect DCC file transfer support by a feature flag
Manuel Nickschas [Fri, 10 Jun 2016 22:10:33 +0000 (00:10 +0200)]
dcc: Add rudimentary view for file transfers
For now this is just a stock QTableView for visualizing the
TransferModel. Will be replaced by a proper view later.
Manuel Nickschas [Wed, 8 Jun 2016 20:18:14 +0000 (22:18 +0200)]
dcc: Introduce TransferModel for DCC file transfers
This model serves as input for a yet-to-come widget for displaying
the current list of transfers.
Manuel Nickschas [Tue, 27 Sep 2016 21:03:30 +0000 (23:03 +0200)]
Use override in a few more places to avoid compiler warnings
Clang complains about inconsistent use of virtual and override, and
for some reason the only offender when building against Qt 5 was
multilineedit.h. This commit fixes the ungodly amount of warnings that
that one header produced.
When building against Qt 4 we still get warnings from the various
application classes; however fixing them triggers a million new
warnings related to Q_OBJECT which is beyond our control, so we'll
just have to wait until we can finally deprecate support for Qt 4
for good...
Manuel Nickschas [Tue, 27 Sep 2016 20:28:46 +0000 (22:28 +0200)]
Avoid deprecation warnings with Qt 5.7
The versioned QStyleOption derivatives are deprecated in Qt 5,
and Qt 5.7 actually gets noisy about that. So remove these by
the unversioned class names.
Michael Marley [Mon, 19 Sep 2016 21:59:21 +0000 (17:59 -0400)]
Fix my AboutData entry
My PPAs only work for Ubuntu, not Debian.
Resolves GH-251.
Manuel Nickschas [Tue, 20 Sep 2016 21:06:10 +0000 (23:06 +0200)]
Update ThanksTo in the AboutDlg
This was pretty outdated and mentioned companies that no longer exist.
Bring the content of that tab into the present. Add new icons, remove
old.
Also modernize the code a tiny bit, and ease the life of translators
by factoring out most of the RichText stuff.
Bernhard Scheirle [Thu, 15 Sep 2016 11:11:34 +0000 (13:11 +0200)]
Updates old links/urls in the about dialog
Manuel Nickschas [Mon, 19 Sep 2016 21:18:39 +0000 (23:18 +0200)]
Update ChangeLog
Manuel Nickschas [Mon, 19 Sep 2016 21:17:50 +0000 (23:17 +0200)]
Update AboutData
New contributors, and one new author!
Hannah von Reth [Thu, 8 Sep 2016 11:30:26 +0000 (13:30 +0200)]
Use Qt::AA_UseHighDpiPixmaps
This will fix some issues... and introduce new ones:
* Some pixmaps are now cut
* Old banner pixmap still pixelated
* Overall looks much less pixelated
Resolves GH-248.
Michael Marley [Wed, 7 Sep 2016 21:34:20 +0000 (17:34 -0400)]
Build with Qt5 by default
This basically reverses the old default behavior. Now, Qt5 is used
by default if neither USE_QT4 nor USE_QT5 is set or if they are
both set.
Resolves GH-246.
Shane Synan [Tue, 6 Sep 2016 09:49:16 +0000 (04:49 -0500)]
Use case-insensitive match for buffer search
When filtering buffers via the quick search bar (Ctrl-S), use
case-insensitive matching. This allows typing lowercase to find
proper names, makes randomly hitting shift not matter, etc. This
mimics the default behavior of the Ctrl-F find bar.
Resolves GH-245.
romibi [Thu, 19 Nov 2015 17:55:00 +0000 (18:55 +0100)]
Enable Nick Toolbar
Nick Toolbar not visible by default.
Resolves GH-243.
Bernhard Scheirle [Fri, 19 Aug 2016 06:06:40 +0000 (08:06 +0200)]
Only load the `EnableSpellCheck` setting for KDE4 users.
Others either have no spell checking at all or use KF5 Sonnet with
its own settings / settings page.
Resolves GH-242.
Bernhard Scheirle [Fri, 19 Aug 2016 06:15:03 +0000 (08:15 +0200)]
Prevent the spelling highlighter from disabling itself
Resolves GH-241.
A. V. Lukyanov [Thu, 11 Aug 2016 07:57:27 +0000 (10:57 +0300)]
Fix multiline textedit not expanding on OS X
Fixes #1212. Resolves GH-240.
A. V. Lukyanov [Wed, 3 Aug 2016 18:47:26 +0000 (21:47 +0300)]
On OS X restore from both dock and tray
A. V. Lukyanov [Wed, 3 Aug 2016 02:16:29 +0000 (05:16 +0300)]
On OS X restore from dock
A. V. Lukyanov [Tue, 2 Aug 2016 18:34:56 +0000 (21:34 +0300)]
On OS X allow minimize on close
Shane Synan [Wed, 27 Jul 2016 06:53:53 +0000 (02:53 -0400)]
QueryBufferItem disconnect IrcUser when removing
Add call to disconnect() when removing IrcUser from QueryBufferItem,
preventing _ircUser from triggering removeIrcUser() again when it's
destroyed by going out of scope. Additionally, limit dataChanged()
to when IrcUser is removed. May offer slight speed benefits.
Fixes at least one case of the client showing some nicknames as
offline when they're actually available (e.g. restarting the client
without restarting the core reveals them as online).
Test case from Bitlbee - receive following (no delay between AWAYs):
:nick!name@gmail.com JOIN :&bitlbee
:nick!name@gmail.com AWAY :Away (Testing)
:nick!name@gmail.com QUIT :Leaving...
:nick!name@gmail.com AWAY :User is offline
Before, 'nick' was wrongly shown as not available. After, 'nick' is
shown as away with the message "User is offline".
(Bitlbee does this as GTalk/XMPP allow for messaging offline users)
Resolves GH-239.
Shane Synan [Tue, 6 Sep 2016 21:33:58 +0000 (16:33 -0500)]
Show IRC server error messages when unexpected
Handle ERROR replies from IRC servers, displaying the error message
in the status buffer.
Add method 'disconnectExpected' to CoreNetwork to expose when a
disconnect is expected. As we're expecting a server error message
when issuing QUIT, hide it to avoid showing redundant information.
This fixes users getting disconnected by the server without any way
to find out why.
Resolves GH-238.
Shane Synan [Tue, 26 Jul 2016 21:42:45 +0000 (17:42 -0400)]
Fix disconnected networks not collapsing on login
Remove call to expandAll() - later calls to setExpandedState() should
automatically expand connected networks.
Before, Qt4 would collapse disconnected networks and expand connected
networks, but Qt5 would always expand both. After, both collapse
disconnected and expand connected.
Resolves GH-237.
Hannah von Reth [Sat, 9 Jul 2016 18:04:59 +0000 (20:04 +0200)]
Use Qt5.7 for appveyor.
Also use gcc 5.3 shipped by Qt.
Resolves GH-231.
romibi [Sun, 28 Aug 2016 12:05:38 +0000 (14:05 +0200)]
Add popup to "(Dis-)Connect from/to IRC" to prevent unwanted action
Resolves GH-160.
romibi [Sun, 14 Feb 2016 10:50:44 +0000 (11:50 +0100)]
Add function to "(Dis-)Connect from/to all" in connection dropdown
If I'm not mistaken there needs to be one ActionType per different
button layout (with icon/text only)...
romibi [Thu, 19 Nov 2015 13:06:01 +0000 (14:06 +0100)]
Change Icons
romibi [Thu, 19 Nov 2015 13:04:56 +0000 (14:04 +0100)]
Add "Connect to Core" and "Disconnect from Core" buttons to main toolbar
Manuel Nickschas [Tue, 13 Sep 2016 22:46:07 +0000 (00:46 +0200)]
Remove CoreInfo from the ClientRegistered handshake message
This string was only used by pre-0.5 clients, and even then only
for displaying information on the connection dialog during the
handshake phase, so it is safe to remove. This prevents the core
from leaking information about itself before a successful login.
Shane Synan [Wed, 7 Sep 2016 05:35:48 +0000 (00:35 -0500)]
Skip rate limit for login, capability negotiation
Disable message rate limit during nick registration and capability
negotiation (Network::Initializing). Once network initialization
is complete (Network::Initialized), re-apply configured rate limits.
Helps fix connection timeouts during login on servers with SASL
enabled, multiple capabilities negotiated, and a strict 10s login
time.
Add flag 'forceUnlimited' to updateRateLimiting() which temporarily
overrides user preferences to skip all rate limiting. User
preferences can be applied afterwards by calling updateRateLimiting()
without setting 'forceUnlimited'.
Resolves GH-236.
Shane Synan [Mon, 5 Sep 2016 21:26:44 +0000 (16:26 -0500)]
Implement custom rate limits
Add custom rate limits flag and settings.
> In database, add 'usecustommessagerate', 'messagerateburstsize',
'messageratedelay', and 'unlimitedmessagerate' to 'network' table,
defaulting to the values already used. This modifies schema,
requiring a version bump, changes to migration scripts, etc.
Use 4 upgrade scripts as SQLite does not support adding multiple
columns at once.
> In protocol, add 'useCustomMessageRate', 'msgRateBurstSize',
'msgRateMessageDelay', and 'unlimitedMessageRate' to Network,
configuring custom rate limits.
Add 'CustomRateLimits' feature flag so the client can tell users
about needing to upgrade their core to get this feature.
If rate limits set to unlimited, disable the token bucket timer and
send messages immediately. Otherwise, modify the existing token
bucket variables according to the new limits. The defaults are
unchanged.
Add "Use Custom Rate Limits" checkbox to network "Connection" tab
with an explanatory tooltip warning against setting too low of rate
limits. Tooltip is modified and checkbox disabled if the core does
not support the feature.
Testing:
> Database schema handling
SQLite new install - works
SQLite upgrade - works
SQLite migrate to Postgres - works
Postgres new install - works
Postgres upgrade - works
> Client/core connections
New client, new core - works
New client, old core - works, custom rates grayed out
Old client, new core - works, custom rates reset + disabled on edit
(Unfortunately there's no way to use a boolean and tell if the client
is old, or actually disabling custom rate limits. Would need the
work-in-progress ClientFeature system 'esainane' designed)
Manuel Nickschas [Tue, 13 Sep 2016 21:44:46 +0000 (23:44 +0200)]
Fix snapcraft version
The previous commit was originally destined for the 0.12 branch, so
the version needs to be adapted.
Christian [Sat, 16 Jul 2016 14:48:10 +0000 (16:48 +0200)]
Snapcraft packaging for quasselcore
Resolves GH-235.
Manuel Nickschas [Wed, 7 Sep 2016 23:40:09 +0000 (01:40 +0200)]
Update INSTALL for WebEngine/WebKit
WebEngine is a new feature, and WebKit is deprecated and unmaintained.
Update INSTALL to reflect the changes.
Manuel Nickschas [Wed, 7 Sep 2016 23:31:19 +0000 (01:31 +0200)]
Make Breeze the default icon theme
Also fix some related UI issues.
romibi [Sat, 16 Jul 2016 13:02:12 +0000 (15:02 +0200)]
Adapt macosx_makePackage.sh for svg icon support
Resolves GH-234.
romibi [Fri, 15 Jul 2016 20:24:32 +0000 (22:24 +0200)]
Enable Breeze on Appveyor
romibi [Fri, 15 Jul 2016 19:37:03 +0000 (21:37 +0200)]
Add Settings option to choose from bundled Icon Themes
or system default
romibi [Fri, 15 Jul 2016 19:35:44 +0000 (21:35 +0200)]
Add and Initialize Icon Themes
romibi [Fri, 15 Jul 2016 19:33:47 +0000 (21:33 +0200)]
Add Icons from Breeze and Breeze Dark
romibi [Fri, 15 Jul 2016 19:31:32 +0000 (21:31 +0200)]
Fix Icon Theme Import Script
Shane Synan [Wed, 13 Jul 2016 14:51:11 +0000 (10:51 -0400)]
Add spacers to Chat & Nick Lists settings page
Add horizontal and vertical spacers to Use Custom Colors and
Custom Nick List toggles on Chat & Nick Lists settings page. Adjust
the grid layout to maintain similar spacing as prior.
This keeps it from stretching across the entire screen when the
settings dialog is very wide (e.g. maximized).
Resolves GH-232.
Shane Synan [Wed, 13 Jul 2016 14:27:37 +0000 (10:27 -0400)]
Add horizontal spacer to Appearance settings page
Add horizontal spacer to Message Redirection toggles on Appearance
settings page. Adjust the grid layout to maintain similar spacing as
prior.
This keeps it from stretching across the entire screen when the
settings dialog is very wide (e.g. maximized).
Shane Synan [Wed, 13 Jul 2016 15:27:43 +0000 (11:27 -0400)]
Auto-resize Settings to fit wide widgets
Add check for settingsTree width. If it's not at the maximum width,
resize the Settings dialog to fit it. Qt should keep the dialog
within bounds of the screen, and the user can always resize the
dialog if needed.
This fixes the Settings dialog squashing the settings tree with the
Sonnet spell-checking widget, Ubuntu 16.04 on en_US, and a 1920×1080
screen.
romibi [Thu, 14 Jul 2016 09:36:07 +0000 (11:36 +0200)]
Enable Qt's high-DPI scaling introduced in Qt5.6
Resolves GH-233.
romibi [Sat, 9 Jul 2016 16:30:44 +0000 (18:30 +0200)]
Fix cmd+Q invalidating the layout somehow
On Mac OSX Qt Applications (with Frameworks packaged) receive the
CloseEvent twice. (See https://bugreports.qt.io/browse/QTBUG-43344)
This triggers a bug where Quassels Main-Window stays hidden and all
Layout states get reset on next launch.
Workaround by checking if event already received...
Resolves GH-230.
romibi [Tue, 5 Jul 2016 19:39:39 +0000 (21:39 +0200)]
Fix Mac Deploy scripts for newer Xcode and git
Resolves GH-229.
romibi [Tue, 5 Jul 2016 19:33:42 +0000 (21:33 +0200)]
Enable OSX dmg deployment on Travis
Resolves GH-228.
Martin T. H. Sandsmark [Sat, 2 Jul 2016 16:24:05 +0000 (18:24 +0200)]
Disable webkit by default
Qt WebKit is deprecated and a bit of a walking security hole, so it
makes sense to disable by default.
Resolves GH-225.
Michael Marley [Sat, 2 Jul 2016 14:20:58 +0000 (10:20 -0400)]
Properly handle 0 seconds in secondsToString()
Previously if 0 seconds was passed in, an empty string would be
returned. Instead, "0 sec" should be returned.
Resolves GH-223.
selabnayr [Fri, 24 Jun 2016 21:38:04 +0000 (14:38 -0700)]
Allow client-side channel buffer merges, and associated changes.
networkmodel.cpp: Mark ChannelBufferItems as being a drop target, to
allow for channel buffer merging.
bufferviewfilter.cpp: Rework BufferViewFilter::flags() to make it more
clear what's going on, and remove the restriction that made
QueryBuffers the only merge-able buffers.
bufferview.cpp: Rework BufferView::dropEvent() to add a bunch of
comments, and allow ChannelBuffers to be merged as well as QueryBuffers.
ChannelBuffers can only be the source for a merge if they are not
currently joined, to prevent UI weirdness of being in a channel but not
having any associated UI elements.
Resolves GH-220.
selabnayr [Fri, 24 Jun 2016 21:20:50 +0000 (14:20 -0700)]
Allow core-side channel buffer merges
Also fix a typo in the warning for a merge that fails the tests.
Rolf Eike Beer [Fri, 1 Jul 2016 17:04:06 +0000 (19:04 +0200)]
Simplify FindExecInfo
* Do not set EXECINFO_INCLUDES to empty it it is NOTFOUND, it is only
used when EXECINFO_FOUND is set, which is the expected behavior
* Simplify checking for the find_path() and find_library() result,
CMake considers a variable false if its value is VARNAME-NOTFOUND
Resolves GH-222.
Rolf Eike Beer [Fri, 1 Jul 2016 16:54:51 +0000 (18:54 +0200)]
Drop QT_PATH from CMake code
If one wants a specific Qt version then simply set QT_QMAKE_EXECUTABLE
to the correct binary. This has less possible side effects.
Rolf Eike Beer [Fri, 1 Jul 2016 16:50:36 +0000 (18:50 +0200)]
Remove arguments of CMake's else() and endif() commands
The need for them is long gone, but they are still permitted. Remove the
remaining instances to make the code more consistent.
Rolf Eike Beer [Fri, 1 Jul 2016 16:47:06 +0000 (18:47 +0200)]
Simplify CMake code