-
Notifications
You must be signed in to change notification settings - Fork 1
issues
For tangram we right now have four areas of issues and ideas we collect:
The version 1.2 currently under development, the version 1.1 in maintenance,
the idea collection backlog', and the heap` of stuff we don't want to forget
but don't actively work on.
1.2 - Currently assigned to: Martin Göllnitz
In webapp.gradle:50 gradle now complains about a runtime definition of a copy-spec instead of a static one.
1.2 - Currently assigned to: Martin Göllnitz
When we now can move ahead to new interfaces for storage, we should again take a look at it with objectify.
1.2 - Currently assigned to: Martin Göllnitz
Hibernate 5.2.6 eliminates the old access to the enhancement class fixed to a javassist based backend. We now face the job to go over to a generic approach which might use javassist or bytebuddy as a backend.
The default now should be bytebuddy.
The instanciation of the classes implementing the stuff which now is just interfaces is pretty unclear. The was no documentation of the really major change found so far.
Martin Goellnitz / Fri, 13 Jan 2017 02:51:17 555e1f5463e8f46dc03b487f216dbcdc77f674e3
refs #617413 update to hibernate 5.2.6 bytebuddy enhancer completed
1.2
The web.xml still reference servlet 2.5 values and some other files may do so also.
1.2
All users from the users.properties should be added to the allowedUsers set.
1.2 - Currently assigned to: Martin Göllnitz
Like DataNucleus JDO with its ClassRepositoryEnhancer morphia can be extended with dynamic means to extend the set of model classes at runtime.
1.2 - Currently assigned to: Martin Göllnitz
Implement CodeResource with Coma means. Discover CodeResource instances in CM repository. Make the mapping configurable/implementable. Add notifiers on changes.
Backlog
Im Moment ist Version 4.4.0 aktuell."
1.2
Many of the supporrted storage APIs cannot rely on query-strings (anymore) to use the full neede selection functionality. We need to change the tangram APIs with breaking changes to support typed query objects, if it is possible at all.
1.2
CodeMirror uses e.g. 5.14 for releases and subsequently 5.14.2 for patchlevels. So the 5.14.0 is missing - but not in all places. In some areas the .0 has to be appended, while the .2 always has to be appended.
1.2
Morphia doesn't support queries of any kind so far since it is missing a query language which the tangram API relies on.
1.2 - Currently assigned to: Martin Göllnitz
To indicate redirection a null model - not an empty mode - is used, while the AbstractTemplateResolver cannot handle such cases. One of the two needs to be changed.
Question: Is the null model an indicator for other parts of the code?
Backlog - Currently assigned to: Martin Göllnitz
Author: Martin Göllnitz
It is unclear right now, why the underlying database layers support such deletions. We rely fully on the referential integrity checks of the storage layer.
1.2
We want to support the most common JSON libraries and representations with generic templatates which output JSON.
Backlog - Currently assigned to: Martin Göllnitz
Right at the moment our tests with CloudFlare show, that there are too few requests which actually can be cached out of the box with Tangram. HTML pages don't receive caching oriented HTTP headers and even CSS have problems, perhaps with the interpretation of the expiry date.
Heap
We'd like to have a solution for String sets, Maps, and for the pac4j AuthenticationClient instances.
Backlog
If a certain PersistenceCapable class definition inherits lists of references from its superclass, those lists cannot be used. (Example: RootTopic subTopics, elements, and relatedContainers)
Backlog
We can now leave behind the old velocity engine, which doesn't seem to be under active development anymore, and go ahead to FreeMarker.
Backlog
Tangram depends on lang (2) through its velocity dependency and lang3 through its pac4j dependency. We tend to use the latter for the future as the single dependency.
Heap
How about view programmed in Groovy or Java - and not only Velocity or JSP as it is the case right now.
Backlog
The default value for the dispatcher path right now is /s - and none of the examples changes it.
We could decide to make this a required parameter so the the GenericLinkFactory can used it as an @Inject @Named mandatory field. In such a case it could be read with any DI Container directly via injection.
Heap
It would be easier to use and configure the system, if @LinkHandler instances were automatically injection into the or registered with the MetaLinkHandler instance.
Backlog
It would be great if we had means to ensure HTTPS usage for the editor - and probably for other areas of the Tangram based application as well.
Backlog
Right now Tangram fully relies on containers separate from the building and packaging of the Tangram system. It might be a good idea to support a self-contained Tangram based application packaging.
We postponed this topic since the gretty plugin brought as a very nice interim solution half the way.
Heap
Through the use of the Tangram.properties file additional values can be injected into the beans prepared by CDI, like it can be accomplished with XML-files for the Spring Framework or the the properties files of dinistiq.
At the moment we were only able to let this take place after the call of the method marked as @PostConstruct which is obviously too late.
Still we don't want eager injections via @Inject for each and every injection point and we don't want to make beans capable of a split or doubled init process.
Heap - Currently assigned to: Martin Göllnitz
This might reduce or easy the custom conversion implementations we did so far.
Heap
http://docs.cksource.com/CKEditor_3.x/Developers_Guide/Java/Integration
Heap
When using the Google App Engine, startup times are a crucial factory. Perhaps not only faster startup but also a custom warm up might help there.