thanks Dirk, I'm going to review the PR in the next days.

Regards,
Tommaso

Il giorno mer 17 gen 2018 alle ore 14:58 Dirk Rudolph <
[email protected]> ha scritto:

> Hey.
>
> I opened a PR for SLING-7364, the missing piece making SCD compatible with
> solr out of the box:
>
> https://github.com/apache/sling-org-apache-sling-distribution-core/pull/6
> <https://github.com/apache/sling-org-apache-sling-distribution-core/pull/6
> >
>
> Beside of that I created an API to create solr documents from resources
> for serialisation here:
>
>
> https://github.com/Buuhuu/sling-content-distribution-solr/tree/master/sling-content-distribution-solr-bundle/src/main/java/com/github/buuhuu/sling/distribution/solr/documents
> <
> https://github.com/Buuhuu/sling-content-distribution-solr/tree/master/sling-content-distribution-solr-bundle/src/main/java/com/github/buuhuu/sling/distribution/solr/documents
> >
>
> With a sample implementation using Adobes AEM Sample “we retail” here:
>
>
> https://github.com/Buuhuu/sling-content-distribution-solr/tree/master/sling-content-distribution-solr-weretail
> <
> https://github.com/Buuhuu/sling-content-distribution-solr/tree/master/sling-content-distribution-solr-weretail
> >
>
> Let me know what you think.
>
> /Dirk
>
> > On 11 Jan 2018, at 15:38, Dirk Rudolph <[email protected]>
> wrote:
> >
> > Many thanks for the positive Feedback.
> >
> > The last thing missing to make it work ootb with solr is SLING-7364. I
> added a some thoughts to it. Option 1) is instead of sending the package
> info as binary header prepended to the request’s body, sending them as http
> headers. As the overall size of http headers is limited this would require
> SLING-3967 to be implemented. Another feasible and probably the more simple
> solutions for now is 3) omitting the package info if the agent is
> configured to so. Wdyt?
> >
> > I will keep working on defining an API for the documents to export of
> solr. From my experience with AEM the following approach comes to my mind:
> >
> > For each Resource:
> >
> > 1. Adapt to SolrFieldsIterator, iterating over SolrField (name, value)
> > 2. If not null, add each SolrField to the document and done
> > 3. If null, adapt to SolrFields, a marker interface that can be used
> together with any (existing) model, that has getters annotated with
> @SolrField.
> > 4. If not null add each @SolrField getter to the document and done
> > 5. If null, log and skip
> >
> > About step 1) Thinking about cq:Page or dam:Asset in AEM, or anything
> Resource that should be exported as an aggregate of child resources. The
> SolrFieldsIterator implementation would traverse and itself adapt each
> resource to SolrFields to do the same as step 3).
> >
> > This would be done as a service so it could be replaced by any other
> implementation, like a simple property=>field mapping.
> >
> > Thoughts?
> >
> >> On 10 Jan 2018, at 10:59, Tommaso Teofili <[email protected]>
> wrote:
> >>
> >> +1
> >>
> >> thanks a lot Dirk for your contributions (4 PRs so far!). This sounds
> like
> >> a very interesting use case for Sling and SCD.
> >>
> >> Regards,
> >> Tommaso
> >>
> >>
> >>
> >> Il giorno mar 9 gen 2018 alle ore 02:22 Daniel Klco <
> [email protected]>
> >> ha scritto:
> >>
> >>> This is a great idea! I could see this as a use case for a number of
> >>> different integrations and an example would be very helpful for anyone
> >>> looking to do this.
> >>>
> >>> On Mon, Jan 8, 2018 at 1:44 PM, Stefan Seifert <[email protected]
> >
> >>> wrote:
> >>>
> >>>> hello dirk.
> >>>>
> >>>> i think such a feature would be very useful. integration of an
> external
> >>>> search engine on a "higher level" than the oak-level search
> integration
> >>> is
> >>>> a common use case in our projects as well. it's important to be quite
> >>>> flexible what is indexed and what not and how it's indexed - for best
> >>> match
> >>>> of the business requirements and the special features of the targeted
> >>>> search engine.
> >>>>
> >>>> stefan
> >>>>
> >>>>> -----Original Message-----
> >>>>> From: Dirk Rudolph [mailto:[email protected]]
> >>>>> Sent: Monday, January 8, 2018 4:45 PM
> >>>>> To: [email protected]
> >>>>> Subject: Sling to Solr integration using Sling Content Distribution
> >>>>>
> >>>>> Hi devs,
> >>>>>
> >>>>> Recently I was evaluating if it is easily possible to integrate Sling
> >>> into
> >>>>> Solr using Sling Content Distributions and made some great progress
> >>> here:
> >>>>>
> >>>>> https://github.com/Buuhuu/sling-content-distribution-solr
> >>>>> <https://github.com/Buuhuu/sling-content-distribution-solr>
> >>>>>
> >>>>> The repository explains the goal, the why and who also giving
> >>> instructions
> >>>>> on how to use it. To sum it up a bit:
> >>>>>
> >>>>> - SCD’s features map perfectly fine to the requirements when
> integrating
> >>>>> into solr
> >>>>> - It enables us to ingest content on business perspective (not
> technical
> >>>>> perspective as with Oak’s internal indexes)
> >>>>> - It’s flexible (thanks for that already :)
> >>>>>
> >>>>> Though there are some things that require a bit of attention to make
> it
> >>>>> work out of the box. I opened a couple of issues for that:
> >>>>>
> >>>>> SLING-7357
> >>>>> SLING-7358
> >>>>> SLING-7359
> >>>>> SLING-7360
> >>>>> SLING-7364
> >>>>>
> >>>>> And made proposals accordingly (not for the last one as I want to
> >>> discuss
> >>>>> that first)
> >>>>>
> >>>>> So my general question is:
> >>>>>
> >>>>> Is integrating Sling into Solr (or potentially any other kind of
> system
> >>>>> that offers APIs to do so) a valid and envisaged use-case for Sling
> >>>> Content
> >>>>> Distribution? And if so would it make sense to implement a module for
> >>> lets
> >>>>> say Solr as example directly in Sling?
> >>>>>
> >>>>> If so I would volunteer to propose something for that but I think
> >>> flexibly
> >>>>> integrating Sling as framework for any kind of content driven web
> >>>>> applications into Solr as enterprise search application would be a
> nice
> >>>>> feature to offer.
> >>>>>
> >>>>> Thanks for any kind of feedback,
> >>>>>
> >>>>> /Dirk
> >>>>
> >>>
> >
>
>

Reply via email to