You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-dev@jackrabbit.apache.org by Angela Schreiber <an...@adobe.com> on 2015/03/26 15:26:04 UTC
[VOTE] retire oak-mk into attic
Dear Oak Team
During initial phase of building a new JCR content repository (OAK),
we introduced a new persistence layer API (called MicroKernel API)
originally implemented by the oak-mk module, which since then has
been part of the Oak project.
In the mean time the overall architecture and design has evolved and
matured and we ended up replacing the original MicroKernel persistence
API by the NodeStore API. Nowadays we only use the MicroKernel API as
wrapper around the NodeStore API but abandoned the original default
implementation.
Our recommended and maintained Oak setup currently includes the following
options for the persistence layer:
- Segment Nodestore (aka Tar-MK)
- Document Nodestore (Mongo, Memory, RDB)
Given these developments in our code base, I came to believe that we no
longer need the original MK implementation.
Since the duplicate persistence API and the mismatch between documentation,
project structure and recommendation used to cause quite some confusion
for people getting started with Oak, and to minimise the maintenance for
the Oak code base on the other hand, I would therefore like to suggest that
we retire the oak-mk module (e.g. moving it to attic) and removing it of
the main oak pom.xml.
The corresponding JIRA component description should only be adjusted to
reflect this but will continue to exist.
So, please case your vote for this proposal.
In case of objection I kindly ask you for some technical description on
why we need to keep oak-mk in the productive oak project and perform
regular releases.
Thanks and kind regards
Angela
Re: [VOTE] retire oak-mk into attic
Posted by Marcel Reutegger <mr...@adobe.com>.
+1
Regards
Marcel
On 26/03/15 15:26, "Angela Schreiber" <an...@adobe.com> wrote:
>Dear Oak Team
>
>During initial phase of building a new JCR content repository (OAK),
>we introduced a new persistence layer API (called MicroKernel API)
>originally implemented by the oak-mk module, which since then has
>been part of the Oak project.
>
>In the mean time the overall architecture and design has evolved and
>matured and we ended up replacing the original MicroKernel persistence
>API by the NodeStore API. Nowadays we only use the MicroKernel API as
>wrapper around the NodeStore API but abandoned the original default
>implementation.
>
>Our recommended and maintained Oak setup currently includes the following
>options for the persistence layer:
>
>- Segment Nodestore (aka Tar-MK)
>- Document Nodestore (Mongo, Memory, RDB)
>
>Given these developments in our code base, I came to believe that we no
>longer need the original MK implementation.
>
>Since the duplicate persistence API and the mismatch between
>documentation,
>project structure and recommendation used to cause quite some confusion
>for people getting started with Oak, and to minimise the maintenance for
>the Oak code base on the other hand, I would therefore like to suggest
>that
>we retire the oak-mk module (e.g. moving it to attic) and removing it of
>the main oak pom.xml.
>
>The corresponding JIRA component description should only be adjusted to
>reflect this but will continue to exist.
>
>So, please case your vote for this proposal.
>
>In case of objection I kindly ask you for some technical description on
>why we need to keep oak-mk in the productive oak project and perform
>regular releases.
>
>Thanks and kind regards
>Angela
>
>
>
>
Re: [VOTE] retire oak-mk into attic
Posted by Michael Dürig <md...@apache.org>.
+1
Michael
On 26.3.15 3:26 , Angela Schreiber wrote:
> Dear Oak Team
>
> During initial phase of building a new JCR content repository (OAK),
> we introduced a new persistence layer API (called MicroKernel API)
> originally implemented by the oak-mk module, which since then has
> been part of the Oak project.
>
> In the mean time the overall architecture and design has evolved and
> matured and we ended up replacing the original MicroKernel persistence
> API by the NodeStore API. Nowadays we only use the MicroKernel API as
> wrapper around the NodeStore API but abandoned the original default
> implementation.
>
> Our recommended and maintained Oak setup currently includes the following
> options for the persistence layer:
>
> - Segment Nodestore (aka Tar-MK)
> - Document Nodestore (Mongo, Memory, RDB)
>
> Given these developments in our code base, I came to believe that we no
> longer need the original MK implementation.
>
> Since the duplicate persistence API and the mismatch between documentation,
> project structure and recommendation used to cause quite some confusion
> for people getting started with Oak, and to minimise the maintenance for
> the Oak code base on the other hand, I would therefore like to suggest that
> we retire the oak-mk module (e.g. moving it to attic) and removing it of
> the main oak pom.xml.
>
> The corresponding JIRA component description should only be adjusted to
> reflect this but will continue to exist.
>
> So, please case your vote for this proposal.
>
> In case of objection I kindly ask you for some technical description on
> why we need to keep oak-mk in the productive oak project and perform
> regular releases.
>
> Thanks and kind regards
> Angela
>
>
>
>
Re: [VOTE] retire oak-mk into attic
Posted by Thomas Mueller <mu...@adobe.com>.
+1 for moving oak-mk and oak-mk-api to attic and removing all other
MicroKernel implementations
Re: [VOTE] retire oak-mk into attic
Posted by Stefan Guggisberg <st...@gmail.com>.
On Thu, Mar 26, 2015 at 3:26 PM, Angela Schreiber <an...@adobe.com> wrote:
> Dear Oak Team
>
> During initial phase of building a new JCR content repository (OAK),
> we introduced a new persistence layer API (called MicroKernel API)
> originally implemented by the oak-mk module, which since then has
> been part of the Oak project.
>
> In the mean time the overall architecture and design has evolved and
> matured and we ended up replacing the original MicroKernel persistence
> API by the NodeStore API. Nowadays we only use the MicroKernel API as
> wrapper around the NodeStore API but abandoned the original default
> implementation.
>
> Our recommended and maintained Oak setup currently includes the following
> options for the persistence layer:
>
> - Segment Nodestore (aka Tar-MK)
> - Document Nodestore (Mongo, Memory, RDB)
>
> Given these developments in our code base, I came to believe that we no
> longer need the original MK implementation.
oak-mk is the reference implementation for the MicroKernel API (oak-mk-api).
there exist other MicroKernel API implementations in the oak code base
and having a reference implementation and integration tests makes IMO
sense for an API with multiple implementations.
OTOH since the MicroKernel API has been replaced by the NodeStore API
i guess we can remove oak-mk-api, oak-it-mk and all other MicroKernel
implementations aswell.
so,
-1 for removing oak-mk but keeping oak-mk-api
+1 for removing oak-mk, oak-mk-api and all MicroKernel implementations
cheers
stefan
>
> Since the duplicate persistence API and the mismatch between documentation,
> project structure and recommendation used to cause quite some confusion
> for people getting started with Oak, and to minimise the maintenance for
> the Oak code base on the other hand, I would therefore like to suggest that
> we retire the oak-mk module (e.g. moving it to attic) and removing it of
> the main oak pom.xml.
>
> The corresponding JIRA component description should only be adjusted to
> reflect this but will continue to exist.
>
> So, please case your vote for this proposal.
>
> In case of objection I kindly ask you for some technical description on
> why we need to keep oak-mk in the productive oak project and perform
> regular releases.
>
> Thanks and kind regards
> Angela
>
>
>
>