aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2011-01-04lcs, properly-block feature: make it possible for the plugin to fail in ↵Marja Hassinen
blockUntilReady.
2010-12-21lcs: Simplifying the contract between contextkit and plugins.Marja Hassinen
Emitting ready() when the plugin is already in ready state has no effect. Previously this caused the upper layer to resubscribe. Now, to cause that effect, emit failed() and then ready(). No plugins rely on the old behaviour, afaik.
2010-12-20lcs: renamings related to the waitForSubscription.Marja Hassinen
The API addition is now: ContextProperty::waitForSubscription(bool block).
2010-12-20lca, waitForSubscriptionAndBlock: making added iface funcs pure virtual.Marja Hassinen
The new contract between plugin & contextkit is: when we call waitUntilReadyAndBlock() on a plugin, it blocks for as long time as needed and then emits the ready() signal. So, the empty implementation doesn't make sense (no signal would be emitted).
2010-12-20lcs, waitForSubscriptionAndBlock: Streamlining tests.Marja Hassinen
2010-12-20lcs: waitForFinishedAndBlock: fix the "double subscribe" problem.Marja Hassinen
2010-12-20lcs: waitForSubscriptionAndBlock, packaging the tests.Marja Hassinen
2010-12-20lcs: waitForSubscriptionAndBlock, tests.Marja Hassinen
2010-12-20lcs: fixing the waitForSubscriptionAndBlock.Marja Hassinen
Oops, the previous commit contained the first part of the "add waitUntilReadyAndBlock" change. This is the second part.
2010-12-20lcs: fixing the waitForSubscriptionAndBlock.Marja Hassinen
This "plugin constructed" was not meant to be shared between providers.
2010-12-17lcs: fixing the waitForSubscriptionAndBlock.Marja Hassinen
The subscription wasn't really happening (only scheduled), but now it is. This version still suffers from a "multiple subscribe" problem; the Subscribe D-Bus call is sent twice.
2010-12-17lcs: Fixing waitForSubscriptionAndBlock.Marja Hassinen
Some things were queued to "happen the next time we enter the main loop". When waitForSubscriptionAndBlock is called, we need them to happen now. There is still a problem that getting the new value is delayed, so it's not present when the wait returns.
2010-12-17lcs API addition: ContextProperty::waitForSubscriptionAndBlock.Marja Hassinen
This function doesn't spin the event loop but blocks until all the pending subscriptions are processed. This also changes the plugin API, so all our plugins need to be updated. "How to wait and block" is also plugin-dependent, so each plugin must implement this functionality.
2010-12-08Prepare 0.5.25.Marja Hassinen
2010-12-08Release 0.5.24.0.5.24Marja Hassinen
2010-12-08Changelog.Marja Hassinen
2010-12-07provider: "Multiple subscribe" and "unsubscribe without subscribe" are no ↵boottimewarningsMarja Hassinen
longer errors. Sending these messages does no actual harm; the intention of the client is clear. "Multiple subscribe" can actually happen when the provider is not running when the subscriber starts. Contextkit recovers from that situation, and treating "multiple subscribe" as an error provides little value.
2010-12-07Don't print warnings when the provider is not running.Marja Hassinen
This can happen e.g., during boot, and contextkit recovers from it when the provider starts.
2010-12-07Adding comments about some "unexpected but possible" lcs behavior.Marja Hassinen
2010-11-10Prepare 0.5.24Marja Hassinen
2010-11-10Release 0.5.230.5.23Marja Hassinen
2010-11-10Changelog.Marja Hassinen
2010-11-10Merge branch 'fix_c_api'Marja Hassinen
2010-11-09Prepare 0.5.23Jean-Luc Lamadon
2010-11-09Release 0.5.22release_0.5.220.5.22Jean-Luc Lamadon
2010-11-09Add Position.IsFlat of type booleanJean-Luc Lamadon
2010-11-09libcontextprovider, c api: don't restart services when new keys are added.Marja Hassinen
2010-11-09Fixing some debug/warning inconsistency.Marja Hassinen
2010-11-08Prepare 0.5.22.Marja Hassinen
2010-11-08Release 0.5.210.5.21Marja Hassinen
2010-11-08Merge remote branch 'remotes/origin/conditional_commanding'Marja Hassinen
Conflicts: configure.ac debian/changelog
2010-11-03Prepare 0.5.19.3Jean-Luc Lamadon
2010-11-02Release 0.5.19.20.5.19.2Jean-Luc Lamadon
2010-11-02Fix: don't wake up if a ValueChanged signal comes for an unsubscribed property.Marja Hassinen
2010-11-02Added System.OfflineMode to core propertiesJean-Luc Lamadon
2010-11-02Renamed context-intro.txt to index.txt for proper documentationJean-Luc Lamadon
2010-10-18Prepare 0.5.19.2Marja Hassinen
But release 0.5.21 if no more parallel releases are needed.
2010-10-18Release 0.5.19.10.5.19.1Marja Hassinen
(Transition phase for integrating 0.5.20 is still in progress, so this is a parallel release, and in the future the changes for this and the changes for 0.5.20 need to be released together.)
2010-10-18Libcontextsubscriber fix: don't crash if QDBusConnection::interface() == 0.Marja Hassinen
2010-10-15Release 0.5.20release_0.5.200.5.20Jean-Luc Lamadon
2010-10-13Connecting to commander only if CONTEXT_COMMANDING is set.Marja Hassinen
This avoids one extra D-Bus connection if the session bus is not otherwise needed. Also, waitForSubscription is quicker if the value is available: we don't need to wait for the "is commander present" check to complete.
2010-09-27AUTHORSMarja Hassinen
2010-09-21QObject has a virtual dtor, InfoBackend doesn't need one.Marja Hassinen
Revert "Adding a virtual dtor to InfoBackend." This reverts commit daaf9dd2237ac6c97fad25215df4b35e8fb743de.
2010-09-21Adding a virtual dtor to InfoBackend.Marja Hassinen
2010-09-21Prepare 0.5.20Marja Hassinen
2010-09-20Release 0.5.190.5.19Marja Hassinen
2010-09-20changelogMarja Hassinen
2010-09-20Hiding the debug messages in libcontextsubscriber and libcontextprovider.Marja Hassinen
... as proposed in bug 192678. Now the debug prints won't be enabled even when setting the env var CONTEXT_LOG_VERBOSITY=debug.
2010-09-20changelogMarja Hassinen
2010-09-20libcontextsubscriber fix: if plugin loading failed, waitForSubscription hanged.Marja Hassinen
generated by cgit v1.2.3 (git 2.25.1) at 2024-07-06 01:32:16 +0000