You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sling.apache.org by Timothee Maret <tm...@apache.org> on 2017/01/20 14:47:16 UTC
Integration tests with specific Oak backend
Hi,
I would like to add an integration test that covers SCD binary less feature
(Use Binary References) in a specific setup:
Two instances with shared caching blob store (AFAIK any blob
store that implements [0] such as [1]).
I wonder two things
1. Is it fine to run integration tests using FileDataStores (rather than
the potentially much faster memory based data stores) ?
2. Would it make sense to complete the
org.apache.sling.testing.sling-mock-oak in order to allow specifying what
backend configuration to use ?
Regards,
Timothee
[0] org.apache.jackrabbit.oak.plugins.blob.AbstractSharedCachingDataStore
[1] org.apache.jackrabbit.oak.plugins.blob.datastore.CachingFileDataStore
Re: Integration tests with specific Oak backend
Posted by Timothée Maret <ti...@gmail.com>.
Hi Oliver,
2017-01-23 15:58 GMT+01:00 Oliver Lietz <ap...@oliverlietz.de>:
> On Monday 23 January 2017 15:33:58 Timothee Maret wrote:
> > Hi,
> >
> > 2017-01-20 18:15 GMT+01:00 Oliver Lietz <ap...@oliverlietz.de>:
> > > On Friday 20 January 2017 15:47:16 Timothee Maret wrote:
> > > > Hi,
> > >
> > > Hi,
> > >
> > > > I would like to add an integration test that covers SCD binary less
> > >
> > > feature
> > >
> > > > (Use Binary References) in a specific setup:
> > > > Two instances with shared caching blob store (AFAIK any
> blob
> > > >
> > > > store that implements [0] such as [1]).
> > > >
> > > > I wonder two things
> > > >
> > > > 1. Is it fine to run integration tests using FileDataStores (rather
> than
> > > > the potentially much faster memory based data stores) ?
> > >
> > > Oak Server, Scripting Thymeleaf and FreeMarker use "real" Sling
> instances
> > > with
> > > SegmentNodeStore already and Sling Karaf a DocumentNodeStore with Mongo
> > > also –
> > > so it should be fine. See Michael's comment in OAK-4862 and be careful
> > > when
> > > using mocks and MemoryNodeStore.
> >
> > Thanks for the pointers!
> >
> > > > 2. Would it make sense to complete the
> > > > org.apache.sling.testing.sling-mock-oak in order to allow specifying
> > >
> > > what
> > >
> > > > backend configuration to use ?
> > >
> > > Is it then still a mock?
> >
> > I thought this was debatable. I assume the consensus is to not provide
> > realistic mocks, and only setup specific backend if really needed.
>
> Completing sling-mock-oak makes sense indeed as there are already "mocks"
> based on real repositories.
>
+1 I have opened SLING-6483 to track this.
Regards,
Timothee
>
> O.
>
> >
> > Regards,
> >
> > Timothee
> >
> > > Regards,
> > > O.
> > >
> > > > Regards,
> > > >
> > > > Timothee
> > > >
> > > > [0] org.apache.jackrabbit.oak.plugins.blob.
> > >
> > > AbstractSharedCachingDataStore
> > >
> > > > [1] org.apache.jackrabbit.oak.plugins.blob.datastore.
> > >
> > > CachingFileDataStore
>
>
--
Timothée Maret
Re: Integration tests with specific Oak backend
Posted by Oliver Lietz <ap...@oliverlietz.de>.
On Monday 23 January 2017 15:33:58 Timothee Maret wrote:
> Hi,
>
> 2017-01-20 18:15 GMT+01:00 Oliver Lietz <ap...@oliverlietz.de>:
> > On Friday 20 January 2017 15:47:16 Timothee Maret wrote:
> > > Hi,
> >
> > Hi,
> >
> > > I would like to add an integration test that covers SCD binary less
> >
> > feature
> >
> > > (Use Binary References) in a specific setup:
> > > Two instances with shared caching blob store (AFAIK any blob
> > >
> > > store that implements [0] such as [1]).
> > >
> > > I wonder two things
> > >
> > > 1. Is it fine to run integration tests using FileDataStores (rather than
> > > the potentially much faster memory based data stores) ?
> >
> > Oak Server, Scripting Thymeleaf and FreeMarker use "real" Sling instances
> > with
> > SegmentNodeStore already and Sling Karaf a DocumentNodeStore with Mongo
> > also –
> > so it should be fine. See Michael's comment in OAK-4862 and be careful
> > when
> > using mocks and MemoryNodeStore.
>
> Thanks for the pointers!
>
> > > 2. Would it make sense to complete the
> > > org.apache.sling.testing.sling-mock-oak in order to allow specifying
> >
> > what
> >
> > > backend configuration to use ?
> >
> > Is it then still a mock?
>
> I thought this was debatable. I assume the consensus is to not provide
> realistic mocks, and only setup specific backend if really needed.
Completing sling-mock-oak makes sense indeed as there are already "mocks"
based on real repositories.
O.
>
> Regards,
>
> Timothee
>
> > Regards,
> > O.
> >
> > > Regards,
> > >
> > > Timothee
> > >
> > > [0] org.apache.jackrabbit.oak.plugins.blob.
> >
> > AbstractSharedCachingDataStore
> >
> > > [1] org.apache.jackrabbit.oak.plugins.blob.datastore.
> >
> > CachingFileDataStore
Re: Integration tests with specific Oak backend
Posted by Timothee Maret <tm...@apache.org>.
Hi,
2017-01-20 18:15 GMT+01:00 Oliver Lietz <ap...@oliverlietz.de>:
> On Friday 20 January 2017 15:47:16 Timothee Maret wrote:
> > Hi,
>
> Hi,
>
> > I would like to add an integration test that covers SCD binary less
> feature
> > (Use Binary References) in a specific setup:
> >
> > Two instances with shared caching blob store (AFAIK any blob
> > store that implements [0] such as [1]).
> >
> > I wonder two things
> >
> > 1. Is it fine to run integration tests using FileDataStores (rather than
> > the potentially much faster memory based data stores) ?
>
> Oak Server, Scripting Thymeleaf and FreeMarker use "real" Sling instances
> with
> SegmentNodeStore already and Sling Karaf a DocumentNodeStore with Mongo
> also –
> so it should be fine. See Michael's comment in OAK-4862 and be careful when
> using mocks and MemoryNodeStore.
>
>
Thanks for the pointers!
> > 2. Would it make sense to complete the
> > org.apache.sling.testing.sling-mock-oak in order to allow specifying
> what
> > backend configuration to use ?
>
> Is it then still a mock?
>
I thought this was debatable. I assume the consensus is to not provide
realistic mocks, and only setup specific backend if really needed.
Regards,
Timothee
>
> Regards,
> O.
>
> > Regards,
> >
> > Timothee
> >
> > [0] org.apache.jackrabbit.oak.plugins.blob.
> AbstractSharedCachingDataStore
> > [1] org.apache.jackrabbit.oak.plugins.blob.datastore.
> CachingFileDataStore
>
>
Re: Integration tests with specific Oak backend
Posted by Oliver Lietz <ap...@oliverlietz.de>.
On Friday 20 January 2017 15:47:16 Timothee Maret wrote:
> Hi,
Hi,
> I would like to add an integration test that covers SCD binary less feature
> (Use Binary References) in a specific setup:
>
> Two instances with shared caching blob store (AFAIK any blob
> store that implements [0] such as [1]).
>
> I wonder two things
>
> 1. Is it fine to run integration tests using FileDataStores (rather than
> the potentially much faster memory based data stores) ?
Oak Server, Scripting Thymeleaf and FreeMarker use "real" Sling instances with
SegmentNodeStore already and Sling Karaf a DocumentNodeStore with Mongo also –
so it should be fine. See Michael's comment in OAK-4862 and be careful when
using mocks and MemoryNodeStore.
> 2. Would it make sense to complete the
> org.apache.sling.testing.sling-mock-oak in order to allow specifying what
> backend configuration to use ?
Is it then still a mock?
Regards,
O.
> Regards,
>
> Timothee
>
> [0] org.apache.jackrabbit.oak.plugins.blob.AbstractSharedCachingDataStore
> [1] org.apache.jackrabbit.oak.plugins.blob.datastore.CachingFileDataStore