Why not just use the StAX API’s built into the JDK to avoid another dependency?
That said, it’s likely using StAX might/should bring in woodstox.
Dan
> On Aug 9, 2017, at 5:11 AM, Jean-Baptiste Onofré wrote:
>
> Fair enough. Agree, worth to evaluate xmlpullparser.
>
> Regards
> JB
>
> O
Fair enough. Agree, worth to evaluate xmlpullparser.
Regards
JB
On 08/09/2017 11:04 AM, Fabian Lange wrote:
I cannot really tell.
Whatever XML stream or Document parser we would use, it also would load new
classes and require some time to initialize.
JAXB is however known to be the most ineffi
I cannot really tell.
Whatever XML stream or Document parser we would use, it also would load new
classes and require some time to initialize.
JAXB is however known to be the most inefficient in that regard. On the
other hand it offers the most convenience because you get these nice mapped
POJOs.
Hi Fabian,
interesting fact indeed. What alternate do you propose ?
Regards
JB
On 08/09/2017 10:46 AM, Fabian Lange wrote:
Hey,
I have new intel on this. A colleague generated a flame graph.
I cannot share the raw svg unfortunately, but I cropped the Feature JAXB
parsing.
It is about 25% of
Hey,
I have new intel on this. A colleague generated a flame graph.
I cannot share the raw svg unfortunately, but I cropped the Feature JAXB
parsing.
It is about 25% of the overall startup time just to boot up the jaxb
infrastructure.
Fabian
On Fri, Jul 22, 2016 at 10:06 AM, Fabian Lange
wrote
Note that on older branches, the feature repositories are still parsed
using DOM
https://github.com/apache/karaf/blob/karaf-2.x/features/core/src/main/java/org/apache/karaf/features/internal/RepositoryImpl.java
The xml is now also written, that may be the reason why we switched.
2016-07-14 23:11
It's not a terribly complex data model, doing it with StAx by hand
wouldn't look so bad.
On Thu, Jul 14, 2016 at 5:11 PM, Fabian Lange
wrote:
> Hi,
>
> i am looking into ways to trim down Karaf. I notices that Karaf uses JaxB
> to parse features.xml - However this is the only application of JaxB.
Hi Fabian,
clearly, we adopted JAXB because we directly created the annotated POJO
and the marshalling/unmarshalling (as the feature parsing is also used
in the maven plugin MOJO) and it was easier than "manual" parsing.
Maintenance is also easier as just updated the POJOs work.
What would y
