Bartosz Małkowski opened 1 decade ago
|
|
Sample call order:
|
|
Bartosz, you are or you will be working on a new configuration API/framework so I am assigning this task to you. However, we won't be able to do it for 7.0 release anyway, so pushed back for next version. |
|
Bartosz what is the status of this feature? If it is not completed yet within the new component framework but you still plan to do it, please provide work estimation. |
|
It is integrated with Kernel-Based-TCF. WIll be merged with branch master later. |
|
Do we really want to introduce this into 7.1.0? Maybe we should start working concurrently on version 7.2.0? I mean, we approach to closing version 7.1.0 as soon as possible but at the same time we continue on pushing all the new stuff to the version 7.2.0 and make it new official development version? What do you think: %bmalkow , %wojtek , %andrzej.wojcik , %eric , %daniel |
|
Two comments:
This way it could allow us to continue with the regular development.
Of course it would be nice if the new TCF-Kernel-Config-etc has been documented (javadoc, asciidoc in the Developer Guide) before release/inclusion :) (comment made private, not sure if we should discuss release related matters publicly) |
|
Wojciech Kapcia wrote:
I am keen on following or adopting workflow from other projects it that makes sense to us, so I have no objections in changing it, if it makes our life easier and also is easier to understand for people who use our code. And I guess adopting a workflow which is already publicly known makes it easier for our users to follow it. I am worried about these 2 things:
|
|
I hope Kernel-based-TCF and converted components should works as current version. But I can't guarantee it. Do not put such big changes to production code without tests, please! |
|
I think we might want to push this to 7.2.0 and fully test it before release, we recently had a major issue (and will likely have more once we release 7.1.0) with the PubSub Schema changes, we should limit the number of major changes to code and operation to keep service interruption with our clients to a minimum. Alternatively we could introduce it in a smaller 7.2.0 update and bump up the release schedule to publish it sooner. |
|
Agreed. After Bartosz comments I am even more convinced that we should not put this new code into 7.1.0.
Let's make the 7.2.0 our official development branch and 7.1.0 release branch or whatever we call it. I will walk through all the 7.1.0 open tickets and review them once again to make sure we can finally close 7.1.0 version. |
|
Artur Hefczyc wrote:
I've create new branch |
|
Ok, good, thank you Wojciech. Bartek, in such a case, please merge all your recent new code and features, frameworks and APIs to the master branch, so we can start working on the new code and testing it. I would like the code to be deployed to tigase.org and tigase.im as soon as possible. |
|
Code of ligase-server is merged already. What about tigase-muc and tigase-pubsub? Should %wojtek make similar branches there? |
|
Appropriate branches created. Due to recent failed builds I've bumped tigase-server version in |
|
Artur, one more comment - there was also potential issue with generating nightlies so I've modified the jenkins job so it will build nightly from both master and, if exists, release branch - this way it will be possible to have subsequent releases of currently-under-release version next to current regular development versions from master. However there is still problem with javadocs of new code that would cause the build to fail, e.g.:
|
|
Bartek, I guess the JavaDoc errors come from your code, please fix it, so we can get a clean build. |
|
I've corrected javadocs on Friday, however there are still problems with two sets of automatic tests.
Then there is the case with one custom component (User Search) which uses TCF. |
|
reported in #3611 |
|
What is progress of this? |
|
Problems reported (in #3611) by Wojtek, ale fixed. I implemented configuring each bean in Kernel by ad-hoc command or by init.properties file. Unfortunatelly, I didn't changed a way in what Tigase server properties. I think it should be done during migrating to Kernel. |
|
Ok, so it goes to version 7.2.0. |
Type |
Task
|
Priority |
Major
|
Assignee | |
RedmineID |
898
|
Version |
tigase-server-8.0.0
|
Estimation |
40h
|
Spent time |
96h
|
There are two problems with setProperties() method.
order of calls is wrong: now, at the beginning are few calls with single property and last call sets all properties (this bulk set contains also keys before setted as separated calls)
will be great, if component configuration changes will be atomic: component accepts whole configuration or reject whole incoming set of properties.
With current implementation is difficult to (for example) change database settings: in one setProperties() call will be set 'db-type' and in the second one: db-url. Is very difficult to manage it correctly.