Should I replace Jena Fuseki 1.1 with 1.3.1 or 2.3.1?

[Steinar Bang]

Hi,

I have an installation that is currently using (as near as I can tell)
Jena Fuseki 1.1, which is a bit out of date.

Going forward, should I would it be best to replace it with Jena Fuseki
1.3.1 or with Jena Fuseki 2.3.1?

By best, I'm thinking of
- performance?
- In particular SPARQL query perfomance
- compatibility?
- REST endpoint compatibility? (that may be determined by the SPARQL
protocol...?)
- Directory layout compatibility? (I know this is different)
- TDB compatibility

Thanks!


- Steinar

[Andy Seaborne]

Hopefully, Fuseki2 is compatible with Fuseki1 configurations - it's
supposed to be anyway.

At some point (no plans), Fuseki1 is end-of-life. Currently, it's only
getting maintenance.

Fuseki2 runs as a service much better.

And as a WAR file.

Post by Steinar Bang
Hi,
I have an installation that is currently using (as near as I can tell)
Jena Fuseki 1.1, which is a bit out of date.
Going forward, should I would it be best to replace it with Jena Fuseki
1.3.1 or with Jena Fuseki 2.3.1?
By best, I'm thinking of
- performance?
- In particular SPARQL query perfomance

Makes no difference - same query engine.

Post by Steinar Bang
- compatibility?

There are better ways to configure Fuseki2 but --conf and the other
command line arguments work.

Post by Steinar Bang
- REST endpoint compatibility? (that may be determined by the SPARQL
protocol...?)

Fuseki2 is better at a bit better at compliance than Fuseki1 (more
complete) and supports normal REST operations on the dataset itself.

For query/update Fuseki2 is compatible with Fuseki1.

Post by Steinar Bang
- Directory layout compatibility? (I know this is different)

It does have it's own "run" area. That's part of being more OS-service
oriented.

Post by Steinar Bang
- TDB compatibility

Same TDB.

Post by Steinar Bang
Thanks!
- Steinar

Andy

[Steinar Bang]

Post by Andy Seaborne
Hopefully, Fuseki2 is compatible with Fuseki1 configurations - it's
supposed to be anyway.
At some point (no plans), Fuseki1 is end-of-life. Currently, it's
only getting maintenance.
Fuseki2 runs as a service much better.

Ok, thanks! Compared with what you said inline, Fuseki2 it is!

One point I forgot to mention was that we want SPARQL query results as
JSON-LD and that's possibly Fuseki2 only...? (If so, this definitely
means going for Fuseki2).

Post by Andy Seaborne

Post by Steinar Bang
- Directory layout compatibility? (I know this is different)

It does have it's own "run" area. That's part of being more
OS-service oriented.

As far as I can tell from the existing code, it operates with a TDB
directory with the dataset embedded, while Fuseki2 seems to use the
structure
run/databases/<dataset-name>

Ie. one separate TDB database per dataset...?

Or is it just that the existing code creates a single dataset and names
the directory it resides in "TDB" rather than using the dataset name...?
(The existing code that wraps Fuseki1 _does_ seem to ensure that the
a single dataset is always present and creates it if it isn't)

[Andy Seaborne]

Post by Steinar Bang

Post by Andy Seaborne
Hopefully, Fuseki2 is compatible with Fuseki1 configurations - it's
supposed to be anyway.
At some point (no plans), Fuseki1 is end-of-life. Currently, it's
only getting maintenance.
Fuseki2 runs as a service much better.

Ok, thanks! Compared with what you said inline, Fuseki2 it is!
One point I forgot to mention was that we want SPARQL query results as
JSON-LD and that's possibly Fuseki2 only...? (If so, this definitely
means going for Fuseki2).

Fuseki1 should support JSON-LD - they both use the same parser/write
code and fuseki1 is being rebuilt with each Jena release.

Post by Steinar Bang

Post by Andy Seaborne

Post by Steinar Bang
- Directory layout compatibility? (I know this is different)

It does have it's own "run" area. That's part of being more
OS-service oriented.

As far as I can tell from the existing code,

Your code? Running Fuseki from within a program?

(That does work in Fuseki2 - jut call main - although a proper embedding
setup would be nicer - but IIRC Fuseki1 was no different)

Post by Steinar Bang
it operates with a TDB
directory with the dataset embedded, while Fuseki2 seems to use the
structure
run/databases/<dataset-name>
Ie. one separate TDB database per dataset...?
Or is it just that the existing code creates a single dataset and names
the directory it resides in "TDB" rather than using the dataset name...?
(The existing code that wraps Fuseki1 _does_ seem to ensure that the
a single dataset is always present and creates it if it isn't)

run/databases/ is where Fuseki2 puts TDB databases created via the UI.
A service configuration can point to anywhere on disk.

The command line way to start Fuseki2 is generally compatible with
Fuseki1 to aid migration. Running Fuseki2 as a OS-service is much easier.

Andy

[Don Rolph]

2.3.1 has the following advantages which we have leveraged:

- it cleanly supports persistent databases

- it has a real security module

Post by Steinar Bang
Hi,
I have an installation that is currently using (as near as I can tell)
Jena Fuseki 1.1, which is a bit out of date.
Going forward, should I would it be best to replace it with Jena Fuseki
1.3.1 or with Jena Fuseki 2.3.1?
By best, I'm thinking of
- performance?
- In particular SPARQL query perfomance
- compatibility?
- REST endpoint compatibility? (that may be determined by the SPARQL
protocol...?)
- Directory layout compatibility? (I know this is different)
- TDB compatibility
Thanks!
- Steinar