Guus der Kinderen <guus.der.kinderen@gmail.com>: Author Summary
| Build | Completed | Code commits | Tests |
|---|---|---|---|
| OPENFIRE › NMRB › #1103 | 2 days ago | 447 passed | |
| OPENFIRE › NMDB › #1058 | 2 days ago |
Docker publishing only after all tests succeed.
Previously, only the Smack test outcome was taken into account (as the others were not stable). Smack's output was not very stable either, which made this seemingly an arbitrary choice. The resulting flow diagram might have added more confusion than worked around actual issues? fixes OF-2266: Do not allow clean-ups interfere with test execution
As cleanup starts pretty much at the same time as the test, it's possible that during the test, an earlier attempt gets cleaned up (causing the test to fail). This has been observed on CI with Github Actions. By using a mock of the TaskEngine, the cleanup task never gets executed, which prevents the issue. Add warm-up connection to connectivity tests
For reasons beyond me, one connection (out of many) for the BOSH connectivity test seems to fail - but only on the infrastructure for Github Actions. From the output, it appears to be the _first_ connection that fails. I suspect that this is triggered by a weird scheduling/timing/eventing thingy that is specific to that environment. This commit adds an additional connection attempt, for which the outcome will be ignored. Hopefully by 'warming up' the system under test, we get more stable results. Simplyfiy CI CodeQL workflow
The Github UI gave this warning: > 1 issue was detected with this workflow: git checkout HEAD^2 is no longer necessary. Please remove this step as Code Scanning recommends analyzing the merge commit for best results. This commit removes that step. Identify docker hub pushes for non-released code as 'alpha'
Openfire builds that are not a release typically identify themselves as 'alpha'. The corresponding Docker images could do the same, which would reduce confusion. |
447 passed |
| OPENFIRE › NMMB › #954 | 2 days ago |
Docker publishing only after all tests succeed.
Previously, only the Smack test outcome was taken into account (as the others were not stable). Smack's output was not very stable either, which made this seemingly an arbitrary choice. The resulting flow diagram might have added more confusion than worked around actual issues? Identify docker hub pushes for non-released code as 'alpha'
Openfire builds that are not a release typically identify themselves as 'alpha'. The corresponding Docker images could do the same, which would reduce confusion. Add warm-up connection to connectivity tests
For reasons beyond me, one connection (out of many) for the BOSH connectivity test seems to fail - but only on the infrastructure for Github Actions. From the output, it appears to be the _first_ connection that fails. I suspect that this is triggered by a weird scheduling/timing/eventing thingy that is specific to that environment. This commit adds an additional connection attempt, for which the outcome will be ignored. Hopefully by 'warming up' the system under test, we get more stable results. fixes OF-2266: Do not allow clean-ups interfere with test execution
As cleanup starts pretty much at the same time as the test, it's possible that during the test, an earlier attempt gets cleaned up (causing the test to fail). This has been observed on CI with Github Actions. By using a mock of the TaskEngine, the cleanup task never gets executed, which prevents the issue. Simplyfiy CI CodeQL workflow
The Github UI gave this warning: > 1 issue was detected with this workflow: git checkout HEAD^2 is no longer necessary. Please remove this step as Code Scanning recommends analyzing the merge commit for best results. This commit removes that step. |
447 passed |
| OPENFIRE › NIGHTLYMAVEN › #1548 | 2 days ago |
Identify docker hub pushes for non-released code as 'alpha'
Openfire builds that are not a release typically identify themselves as 'alpha'. The corresponding Docker images could do the same, which would reduce confusion. Add warm-up connection to connectivity tests
For reasons beyond me, one connection (out of many) for the BOSH connectivity test seems to fail - but only on the infrastructure for Github Actions. From the output, it appears to be the _first_ connection that fails. I suspect that this is triggered by a weird scheduling/timing/eventing thingy that is specific to that environment. This commit adds an additional connection attempt, for which the outcome will be ignored. Hopefully by 'warming up' the system under test, we get more stable results. fixes OF-2266: Do not allow clean-ups interfere with test execution
As cleanup starts pretty much at the same time as the test, it's possible that during the test, an earlier attempt gets cleaned up (causing the test to fail). This has been observed on CI with Github Actions. By using a mock of the TaskEngine, the cleanup task never gets executed, which prevents the issue. Simplyfiy CI CodeQL workflow
The Github UI gave this warning: > 1 issue was detected with this workflow: git checkout HEAD^2 is no longer necessary. Please remove this step as Code Scanning recommends analyzing the merge commit for best results. This commit removes that step. Docker publishing only after all tests succeed.
Previously, only the Smack test outcome was taken into account (as the others were not stable). Smack's output was not very stable either, which made this seemingly an arbitrary choice. The resulting flow diagram might have added more confusion than worked around actual issues? |
447 passed |
| OPENFIRE › TRUNK › #2151 | 3 days ago |
Add warm-up connection to connectivity tests
For reasons beyond me, one connection (out of many) for the BOSH connectivity test seems to fail - but only on the infrastructure for Github Actions. From the output, it appears to be the _first_ connection that fails. I suspect that this is triggered by a weird scheduling/timing/eventing thingy that is specific to that environment. This commit adds an additional connection attempt, for which the outcome will be ignored. Hopefully by 'warming up' the system under test, we get more stable results. |
447 passed |
| OPENFIRE › TRUNK › #2150 | 3 days ago |
Identify docker hub pushes for non-released code as 'alpha'
Openfire builds that are not a release typically identify themselves as 'alpha'. The corresponding Docker images could do the same, which would reduce confusion. Docker publishing only after all tests succeed.
Previously, only the Smack test outcome was taken into account (as the others were not stable). Smack's output was not very stable either, which made this seemingly an arbitrary choice. The resulting flow diagram might have added more confusion than worked around actual issues? |
447 passed |
| OPENFIRE › TRUNK › #2149 | 3 days ago |
Simplyfiy CI CodeQL workflow
The Github UI gave this warning: > 1 issue was detected with this workflow: git checkout HEAD^2 is no longer necessary. Please remove this step as Code Scanning recommends analyzing the merge commit for best results. This commit removes that step. |
447 passed |
| OPENFIRE › TRUNK › #2148 | 3 days ago |
fixes OF-2266: Do not allow clean-ups interfere with test execution
As cleanup starts pretty much at the same time as the test, it's possible that during the test, an earlier attempt gets cleaned up (causing the test to fail). This has been observed on CI with Github Actions. By using a mock of the TaskEngine, the cleanup task never gets executed, which prevents the issue. |
447 passed |
| OPENFIRE › NMDB › #1057 | 3 days ago |
OF-2263: Use ListPager in MUC summary admin page
This prevents the page from listing every page number, when there are many pages. Instead of listing every page number between `1` and `35781` (taking up a considerable amount of space on the page), the page now shows what will be a single line in most browsers: Pages: [ 1 2 3 ... 5779 5780 5781 5782 5783 ... 35779 35780 35781 ] OF-2259: Allow pre-load of MUC rooms to be configurable
Up until this commit, Openfire would always, when it started, load in memory all MUC rooms for which activity was recorded in the last X days. On instances with large amounts of rooms, this leads to a significant delay in boot time. This commit makes this pre-loading behavior configurable. The amount of days can be configured (independently from the 'cleanup' duration, which it was previously tied to) but the behavior can also be disabled completely. Using unique values to prevent inter-test interference
By using different values for username and address, there's less of a chance of individual tests affecting the others (like what we've seen happen in Github Actions). OF-2265: Websocket delivery should not depend on XmppSession being present
Although normally, an XmppSession will be available, that won't be the case when, for example, sending an error about the XmppSession not being available. Not doing packet accounting (and resetting an idle time) is preferable to not delivering the message due to a NullPointerException. Expose Openfire Logs when running aioxmpop tests in Github Actions
OF-2262: Take unloaded rooms into account
Openfire can unload rooms that are inactive from memory. Those rooms remain valid, and will be loaded on-demand. This commit ensures that these rooms are: - listed on the admin console - shown in search results - are added to room counts OF-2259: Re-use XML parser when loading messages from database.
When profiling Openfire that was starting up from a database that contained a lot of MUC rooms with a lot of messages (ofMucConversationLog), a hotspot was identified in the code that parses the database content into XMPP messages. The parser that is used was recreated for every message, which added considerable overhead. This commit ensures that the parser that parses database content into messages (as used during the initial load at boot-time) is re-used. OF-2261: Re-word MUC room memory management
The admin console used wording that suggested that MUC rooms were destroyed after a certain amount of inactivity. That's misleading, as they're only unloaded from memory. This commit re-organizes the corresponding admin console page, and improves wording. Remove AIOXMPP retries, as they're expected to fail anyways.
Bump up verbosity of Openfire logging
|
447 passed |
| OPENFIRE › NMRB › #1102 | 3 days ago |
Bump up verbosity of Openfire logging
OF-2259: Re-use XML parser when loading messages from database.
When profiling Openfire that was starting up from a database that contained a lot of MUC rooms with a lot of messages (ofMucConversationLog), a hotspot was identified in the code that parses the database content into XMPP messages. The parser that is used was recreated for every message, which added considerable overhead. This commit ensures that the parser that parses database content into messages (as used during the initial load at boot-time) is re-used. OF-2262: Take unloaded rooms into account
Openfire can unload rooms that are inactive from memory. Those rooms remain valid, and will be loaded on-demand. This commit ensures that these rooms are: - listed on the admin console - shown in search results - are added to room counts OF-2265: Websocket delivery should not depend on XmppSession being present
Although normally, an XmppSession will be available, that won't be the case when, for example, sending an error about the XmppSession not being available. Not doing packet accounting (and resetting an idle time) is preferable to not delivering the message due to a NullPointerException. Remove AIOXMPP retries, as they're expected to fail anyways.
Expose Openfire Logs when running aioxmpop tests in Github Actions
OF-2261: Re-word MUC room memory management
The admin console used wording that suggested that MUC rooms were destroyed after a certain amount of inactivity. That's misleading, as they're only unloaded from memory. This commit re-organizes the corresponding admin console page, and improves wording. OF-2259: Allow pre-load of MUC rooms to be configurable
Up until this commit, Openfire would always, when it started, load in memory all MUC rooms for which activity was recorded in the last X days. On instances with large amounts of rooms, this leads to a significant delay in boot time. This commit makes this pre-loading behavior configurable. The amount of days can be configured (independently from the 'cleanup' duration, which it was previously tied to) but the behavior can also be disabled completely. Using unique values to prevent inter-test interference
By using different values for username and address, there's less of a chance of individual tests affecting the others (like what we've seen happen in Github Actions). OF-2263: Use ListPager in MUC summary admin page
This prevents the page from listing every page number, when there are many pages. Instead of listing every page number between `1` and `35781` (taking up a considerable amount of space on the page), the page now shows what will be a single line in most browsers: Pages: [ 1 2 3 ... 5779 5780 5781 5782 5783 ... 35779 35780 35781 ] |
447 passed |
| Build | Completed | Code commits | Tests |
|---|---|---|---|
| OPENFIRE › NIGHTLYMAVEN › #1547 | 3 days ago |
Bump up verbosity of Openfire logging
Remove AIOXMPP retries, as they're expected to fail anyways.
Expose Openfire Logs when running aioxmpop tests in Github Actions
OF-2259: Re-use XML parser when loading messages from database.
When profiling Openfire that was starting up from a database that contained a lot of MUC rooms with a lot of messages (ofMucConversationLog), a hotspot was identified in the code that parses the database content into XMPP messages. The parser that is used was recreated for every message, which added considerable overhead. This commit ensures that the parser that parses database content into messages (as used during the initial load at boot-time) is re-used. OF-2259: Allow pre-load of MUC rooms to be configurable
Up until this commit, Openfire would always, when it started, load in memory all MUC rooms for which activity was recorded in the last X days. On instances with large amounts of rooms, this leads to a significant delay in boot time. This commit makes this pre-loading behavior configurable. The amount of days can be configured (independently from the 'cleanup' duration, which it was previously tied to) but the behavior can also be disabled completely. OF-2261: Re-word MUC room memory management
The admin console used wording that suggested that MUC rooms were destroyed after a certain amount of inactivity. That's misleading, as they're only unloaded from memory. This commit re-organizes the corresponding admin console page, and improves wording. Using unique values to prevent inter-test interference
By using different values for username and address, there's less of a chance of individual tests affecting the others (like what we've seen happen in Github Actions). OF-2265: Websocket delivery should not depend on XmppSession being present
Although normally, an XmppSession will be available, that won't be the case when, for example, sending an error about the XmppSession not being available. Not doing packet accounting (and resetting an idle time) is preferable to not delivering the message due to a NullPointerException. OF-2263: Use ListPager in MUC summary admin page
This prevents the page from listing every page number, when there are many pages. Instead of listing every page number between `1` and `35781` (taking up a considerable amount of space on the page), the page now shows what will be a single line in most browsers: Pages: [ 1 2 3 ... 5779 5780 5781 5782 5783 ... 35779 35780 35781 ] OF-2262: Take unloaded rooms into account
Openfire can unload rooms that are inactive from memory. Those rooms remain valid, and will be loaded on-demand. This commit ensures that these rooms are: - listed on the admin console - shown in search results - are added to room counts |
1 of 452 failed |
| OPENFIRE › TRUNK › #2135 | 1 week ago |
OF-2243: Version check task should not be busy doing nothing
Instead of scheduling an immediate task that does nothing for X seconds, schedule a task that will start in X seconds, that immediately does something. This prevents using up a thread from the TaskEngine that's not doing anything for a significant amount of time. |
1 of 452 failed |
| OPENFIRE › NIGHTLYMAVEN › #1519 | 1 month ago |
Rebased to current master branch & use configured values, instead of their defaults.
Reducing complexity by replacing member with local variable.
Use formatting functionality provided by logging framework.
Replacing anonymous class with lambda for brevity
|
447 passed |
| OPENFIRE › NIGHTLYMAVEN › #1102 | 1 year ago |
Update .travis.yml
Disabling Smack Integration Tests. |
1 of 401 failed |
| OPENFIRE › NIGHTLYMAC › #2229 | 2 years ago |
OF-1586: Prevent BOSH from spinning all CPUs.
Refactoring BOSH: reduce complexity.
No functional change. This refactoring primarily removes some duplicate arguments being passed around. Increment Maven release version from 4.2.3 to 4.2.4-SNAPSHOT.
OF-1497: Stream Management should count keepalive pings.
Openfire sends XMPP pings to clients that are idle. These pings should be included in the counts that are kept as part of the Stream Management (XEP-0198) implementation. OF-1670: Allow MUC sysadmins to join a password protected room
When a MUC room is configured to require a password upon entry, a sysadmin should be able to join the room (without providing the password). The above will now be the default configuration. It can be configured, through the admin console, on a per conference service basis (which is the same context in which sysadmins are defined). Refactoring BOSH: Separated out responsiblity for XML parsing in new class HttpBindBody.
Refactoring BOSH: event handling to dedicated class.
OF-1667: Expand on BOSH events
This adds a couple of new BOSH-related events, and makes the servlet context for an event available to the event handler. This direct cause for this change is to allow reading from the BOSH request / URL used, to obtain JWT values. OF-1670: Backport must be Java 7 compatible.
This fixes a backport issue. The original code was based on a version of Openfire that is compiled with Java 8. The branch on which this is backported, 4.2, is compiled with Java 7. OF-1497: Re-enable Stream Management (XEP-0198) by default.
To prevent the unresolved bug OF-1497 from affecting new installations, OF-1563 disabled the Stream Management (XEP-0198) feature by default. With OF-1497 fixed, the feature can be re-enabled again. OF-1667: Drop adapter from backport.
As the newly introduced adapter has not made it in the current master, I'm dropping it here, in favor of creating a new class that's immediately deprecated (master is using Java 8, which allows the adapter to be replaced with default method definitions in the interface). |
Testless build |
| OPENFIRE › NMRB › #58 | 2 years ago |
OF-1605: Add support for wildcard in self-signed certificates.
When generating a self-signed certificate, a wildcard SAN will be generated, unless the property cert.wildcard is set to false (it defaults to true). Using a wildcard makes it more likely that future changes / additions will be covered by the pre-existing certificate. OF-1611: Certificate Manager plugin: should not import more than once per iteration.
OF-1608: Facilitate backup of certificate stores.
OF-1609: Certificate Manager plugin: grace period for file system changes should be respected.
Removed unused error condition definition.
OF-1604: Fix SAN pattern matching
OF-1608: Certificate Manager plugin: create backup before applying changes.
OF-1610: Certificate Manager plugin: use optional passphrase for to-be-imported private key.
OF-1607: Certificate Manager plugin: Set minimum size for certificate chain.
When detecting cerificate chain files, look for (by default) files that at least have _more_ than one certificate in them. Certificate Manager plugin: Add Dutch translation.
OF-1599: Stop insisting on having certificates with two different algorithms
This commit will allow Openfire to operate with just one certificate, instead of two, without showing warnings. The commit also replaces some hard-coded values with properties that have the original hard-coded value as a default. OF-1603: Persist and log only if something changed.
OF-1603: Self-signed certs should be regenerated when the existing one does not match all server identities.
The self-signed certificate that's installed in Openfire should be replaced (when the user instructs to do so) when the existing certificate does not include all server identities as subject alternate names (of the DNS type). The existing code checked only if the primary identity was present in the certificate, and ignored the request to replace the certificate if it was. Certificate Manager plugin: improve logging.
OF-1528: Use Java 8 when compiling JSP.
OF-1607: Certificate Manager plugin: certificate chain detection: should contain more than one certificate.
|
250 passed |
| OPENFIRE › NMDB › #32 | 2 years ago |
OF-1595: External Service Discovery plugin, admin console fix for transport type.
Denote release 1.0.1 of the External Service Discovery plugin.
OF-1592: Don't echo back potentially binary data in VCard error.
OF-1592: Be less verbose when VCard updates fail expectedly.
When the VCard provider is read-only, there's no need to log a VCard update attempt to the 'error' level. This commit reduces the log verbosity, and improves the error sent back to the client. OF-1491: Instruct log4j2 to re-read logfile periodically.
|
246 passed |
| OPENFIRE › NMRB › #30 | 2 years ago |
OF-1491: Instruct log4j2 to re-read logfile periodically.
OF-1592: Be less verbose when VCard updates fail expectedly.
When the VCard provider is read-only, there's no need to log a VCard update attempt to the 'error' level. This commit reduces the log verbosity, and improves the error sent back to the client. OF-1595: External Service Discovery plugin, admin console fix for transport type.
Denote release 1.0.1 of the External Service Discovery plugin.
OF-1592: Don't echo back potentially binary data in VCard error.
|
246 passed |
| OPENFIRE › NMDB › #10 | 2 years ago |
Adjusted javadoc to reflect functional behavior or implementation.
OF-1553: Allow MUC room creation to be limited to registered local users.
The existing options that allow an administrator to configure who is allowed to create new group chat rooms should get an additional option: allow all locally registered users to create rooms. This could be used to prevent anonymous users, and users from other XMPP domains, from creating rooms. OF-852: Add XEP-0115 caps in stream feature
This allows peers to skip service discovery upon connect. OF-852: Recognize node representing Openfire software.
When querying the server for its capabilities, a node can be added to the request that references the software of the server. Openfire should recognize such a node. |
246 passed |
| OPENFIRE › NIGHTLYWINDOWS › #2775 | 3 years ago |
OF-546: Restoring content deleted by mistake. (#1063)
OF-546: Drop Apache Ant in favor of Apache Maven. (#1062)
* OF-546: Drop Apache Ant in favor of Apache Maven. This commit removes support for Apache Ant, and replaces (some, not all) references with references to Apache Maven. * OF-546: Remove bundled libraries for plugins. This removes all bundled binary libraries from plugins that can be build with Maven instead of Ant. |
Testless build |
| Build | Completed | Code commits | Tests |
|---|---|---|---|
| OPENFIRE › NIGHTLYMAVEN › #1548 | 2 days ago |
Identify docker hub pushes for non-released code as 'alpha'
Openfire builds that are not a release typically identify themselves as 'alpha'. The corresponding Docker images could do the same, which would reduce confusion. Add warm-up connection to connectivity tests
For reasons beyond me, one connection (out of many) for the BOSH connectivity test seems to fail - but only on the infrastructure for Github Actions. From the output, it appears to be the _first_ connection that fails. I suspect that this is triggered by a weird scheduling/timing/eventing thingy that is specific to that environment. This commit adds an additional connection attempt, for which the outcome will be ignored. Hopefully by 'warming up' the system under test, we get more stable results. fixes OF-2266: Do not allow clean-ups interfere with test execution
As cleanup starts pretty much at the same time as the test, it's possible that during the test, an earlier attempt gets cleaned up (causing the test to fail). This has been observed on CI with Github Actions. By using a mock of the TaskEngine, the cleanup task never gets executed, which prevents the issue. Simplyfiy CI CodeQL workflow
The Github UI gave this warning: > 1 issue was detected with this workflow: git checkout HEAD^2 is no longer necessary. Please remove this step as Code Scanning recommends analyzing the merge commit for best results. This commit removes that step. Docker publishing only after all tests succeed.
Previously, only the Smack test outcome was taken into account (as the others were not stable). Smack's output was not very stable either, which made this seemingly an arbitrary choice. The resulting flow diagram might have added more confusion than worked around actual issues? |
447 passed |
| OPENFIRE › NIGHTLYMAVEN › #1152 | 1 year ago |
Update muc-room-edit-form.jsp
Fix for broken HTML tag. |
426 passed |
| OPENFIRE › NMRB › #640 | 1 year ago |
Update .travis.yml
Add directive that (for DPL v2) will prevent deployment outside of IgniteRealtime's repository. |
397 passed |
| OPENFIRE › NIGHTLYMAVEN › #1097 | 1 year ago |
Update .travis.yml
Add directive that (for DPL v2) will prevent deployment outside of IgniteRealtime's repository. |
397 passed |
| SMACK › NIGHTLY › #1653 | 2 years ago |
More specific Pubsub integration tests.
XEP-0060 prohibits publishing a request that contains an item to a node that is both 'notification-only' and 'transient' (section 7.1.3.6) In commit 8ed872ca639822aeff58eaaa673f953efb0ba6ee the existing pubsub publication test was modified (to resolve a different issue) to operate on a node that's both 'notification-only' and 'transient'. This resulted in a test that should return an error, even though the test implementation didn't expect one. This commit explicitly verifies that publishing an item to such a node causes an error to be returned. A new test is added that verifies that publishing an event notification does succeed. Admin should be disconnected after tests.
The Smack Integration tests can use an admin account to provision accounts that are used by the tests. This admin account uses an XMPP connection to interact with the server-under-test. When the tests are over, this account should be disconnected explicitly, to prevent stream management from keeping it alive longer than it needs to. |
680 passed |
| OPENFIRE › NIGHTLYRPM › #2857 | 3 years ago |
Restoring installer files used by Ant.
|
Testless build |
| SPARK › CI › #81 | 3 years ago |
SPARK-1963: Prevent calculation of unchanged avatar data.
There's no need to recalculate an avatar hash if the avatar didn't change. Preventing this will improve performance. SPARK-1964: Don't use the UI dispatcher thread when not needed.
Actions that don't affect the UI should not be executed in the heavily synchronized dispatcher. It will needlessly delay execution, but also make the dispatcher thread unavailable for tasks that do belong there. Merge pull request #372 from deleolajide/patch-1
Disable meet plugin by default. Merge pull request #371 from Alameyo/GUI_for_mutual_authentication_and_certificate_creation
SPARK-1985 Support for TLS mutual authentication SPARK-1965: Replace emoticon archives with separate files in version control
Instead of adding the zip files to git, this commit adds the unzipped content, and ensures that the Maven build generates the zip files. This way, editing is a lot easier, and changes are better tracked. Merge pull request #376 from Alameyo/create_empty_KeyStores
SPARK-1989 create empty KeyStores SPARK-1962: Use key-based lookups to find correct emoticon.
This replaces an earlier implementation that'd iterate over all emoticons in a pack when searching for a match with a particular text. Merge pull request #351 from Alameyo/extensions_extraction
SPARK-1934, SPARK-1939 Certificate extensions. Merge pull request #368 from deleolajide/plugin-meet-0.4.0
Openfire meetings pluginfor Spark 0.4.0 SPARK-1956: Preference dialog should remember its size.
The recent fix for SPARK-945 already stored the preference dialog size and position, which fixed most of this bug. When that was implemented, a rogue invocation of .pack() was placed after the stored settings were applied. This resized the component again. SPARK-1962: Don't force smilies to lowercase.
The old code was doing "smiley-text to image" lookup by converting everything to lowercase, then doing a case insensitive lookup. My last commit made that lookup case sensitive, which caused lookups for :-D and :-P to fail. This commit remove the to-lowercase conversion of the text that's used in the lookup, fixing those instances again. SPARK-1888: Properly uninstall a plugin
When uninstalling a plugin, all files from the plugin directory should be removed. SPARK-1888: Reformatting code for Plugin Manager
To fix the issue at hand, improving the readability of the code helps a lot. In this commit, I've applied a code formatter (to just the Plugin Manager code) and made some basic changes (added/improved log messages mostly). SPARK-1961: Disable themes that are CPU-hungry.
This commit disables three themes that each add a considerable load to the startup of Spark. SPARK-1972: New 'security' login settings panel.
This commit adds a new 'security' tab to the advanced login settings. Here, security-related settings are made available. Also, this commit makes slight modifications to the other tabs, by making the components be oriented more to the top-left corner (instead of being centered - which looks pretty ugly). Merge pull request #358 from guusdk/SPARK-1972_New_security-login-settings-panel
SPARK-1972: New 'security' login settings panel. SPARK-1976: Make sure error dialog is centered.
This ensures that the error dialog is centered, in relation to it's parent. Also, when expanding the details (which changes the size of the dialog), the dialog is repositioned again (to remain centered). SPARK-1976: Better error messages
This commit: - reworks the login method to remove some redundant code; - applies the MessageDialog-provided error dialog generator to the login method; - refactored the MessageDialog error dialog generator (to allow for toggling of the large stacktrace pane). In all, this should improve the error messages during login. The same technique could be re-used elsewhere. Merge pull request #356 from Alameyo/SSLContext
SPARK-1966, SPARK-1952, SPARK-1969, SPARK-1970, SPARK-1971 and partially SPARK-1968 SPARK-1888: Uninstall a plugin when the GUI instructs so.
This is a fix to what I assume to be a bug. The method, used by the UI part where plugins are managed, is documented as 'remove and uninstall' a plugin. It only removed, but did not uninstall. I've fixed that with this commit. SPARK-1973: Spark should show an indicator when the connection is not encrypted.
|
2 passed |
| OPENFIRE › NIGHTLYWINDOWS › #2558 | 3 years ago | Testless build | |
| OPENFIRE › NIGHTLYPLUGINS › #2522 | 3 years ago | Testless build | |
| OPENFIRE › NIGHTLYBINDIST › #2620 | 3 years ago | Testless build |
For reasons beyond me, one connection (out of many) for the BOSH connectivity test seems to fail - but only on the infrastructure for Github Actions.
From the output, it appears to be the _first_ connection that fails. I suspect that this is triggered by a weird scheduling/timing/eventing thingy that is specific to that environment.
This commit adds an additional connection attempt, for which the outcome will be ignored. Hopefully by 'warming up' the system under test, we get more stable results.
Previously, only the Smack test outcome was taken into account (as the others were not stable). Smack's output was not very stable either, which made this seemingly an arbitrary choice. The resulting flow diagram might have added more confusion than worked around actual issues?
The Github UI gave this warning:
> 1 issue was detected with this workflow: git checkout HEAD^2 is no longer necessary. Please remove this step as Code Scanning recommends analyzing the merge commit for best results.
This commit removes that step.
Openfire builds that are not a release typically identify themselves as 'alpha'. The corresponding Docker images could do the same, which would reduce confusion.
As cleanup starts pretty much at the same time as the test, it's possible that during the test, an earlier attempt gets cleaned up (causing the test to fail). This has been observed on CI with Github Actions.
By using a mock of the TaskEngine, the cleanup task never gets executed, which prevents the issue.