You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@marmotta.apache.org by Gustavo Mora <gm...@gmail.com> on 2017/03/29 01:12:16 UTC

GSOC - Port marmotta RDF4J

Dear,

I'm a student of Computer Science, and I'm working on my application to
GSOC 2017, but I have some questions. BTW, I also compiled the code and
started playing with it. Firstly, there's a plan to migrate because there
are some classes that have replacements; for example, IRI instead of URI,
changes in some interfaces/classes names, use of optional, etc. Also, some
tests will fail because RDF4J stores a default string as literal, for
example. If we changed that, the next release is not going to have
compatibility with the previous ones? If yes, Is it necessary to maintain
some deprecated code? What other problems can I encounter in KiWi and
platform?

Regards,
Gustavo Mora.

Re: GSOC - Port marmotta RDF4J

Posted by Sergio Fernández <wi...@apache.org>.
Thanks for your interest, Gustavo.

Your proposal looks fine. I've made minor comments on the document.

On Fri, Mar 31, 2017 at 4:22 AM, Gustavo Mora <gm...@gmail.com> wrote:

> Thanks Peter for your answer, I really appreciate, I considered some things
> you mentioned into my proposal. I'd really appreciate if you can review it
> and share some thoughts in order to improve it.
>
> https://docs.google.com/document/d/1oql-LqxpAeTXHVp_
> 2GEj-69yzKVkw0q_LuOoP7aVG9s/edit?usp=sharing
>
> Regards,
> Gustavo.
>
> 2017-03-28 21:28 GMT-05:00 Peter Ansell <an...@gmail.com>:
>
> > Hi Gustavo,
> >
> > One major compatibility issue will be the use of
> > "if(Literal.getLanguage() != null)" in any current code, and if a
> > language tag exists, "Literal.getDatatype()" will now return
> > rdf:langString so you need to make sure that you check for
> > "getLanguage().isPresent()" before checking datatypes. You will be
> > able to pick up most of the getLanguage() != null cases by checking
> > compiler errors.
> >
> > The other incompatibility will be that because Literal.getDatatype no
> > longer ever returns null, you will need to work on a database
> > migration strategy for existing Literal instances that have null, and
> > replace them with xsd:string if the language is missing, and
> > rdf:langString if the language exists.
> >
> > URI and IRI are interchangeable in terms of RDF4J.
> >
> > A fuller migration guide is at:
> >
> > http://docs.rdf4j.org/migration/
> >
> > You may want to split up the migration into steps, by stepping up to
> > Sesame-2.8 first to pick up the getDatatype() != null cases first,
> > then go to Sesame-4 to pick up the getLanguage() != null cases, before
> > going to RDF4J with what are virtually all package name changes.
> >
> > Cheers,
> >
> > Peter
> >
> >
> > On 29 March 2017 at 12:12, Gustavo Mora <gm...@gmail.com> wrote:
> > > Dear,
> > >
> > > I'm a student of Computer Science, and I'm working on my application to
> > > GSOC 2017, but I have some questions. BTW, I also compiled the code and
> > > started playing with it. Firstly, there's a plan to migrate because
> there
> > > are some classes that have replacements; for example, IRI instead of
> URI,
> > > changes in some interfaces/classes names, use of optional, etc. Also,
> > some
> > > tests will fail because RDF4J stores a default string as literal, for
> > > example. If we changed that, the next release is not going to have
> > > compatibility with the previous ones? If yes, Is it necessary to
> maintain
> > > some deprecated code? What other problems can I encounter in KiWi and
> > > platform?
> > >
> > > Regards,
> > > Gustavo Mora.
> >
>



-- 
Sergio Fernández
Partner Technology Manager
Redlink GmbH
m: +43 6602747925
e: sergio.fernandez@redlink.co
w: http://redlink.co

Re: GSOC - Port marmotta RDF4J

Posted by Gustavo Mora <gm...@gmail.com>.
Thanks Peter for your answer, I really appreciate, I considered some things
you mentioned into my proposal. I'd really appreciate if you can review it
and share some thoughts in order to improve it.

https://docs.google.com/document/d/1oql-LqxpAeTXHVp_2GEj-69yzKVkw0q_LuOoP7aVG9s/edit?usp=sharing

Regards,
Gustavo.

2017-03-28 21:28 GMT-05:00 Peter Ansell <an...@gmail.com>:

> Hi Gustavo,
>
> One major compatibility issue will be the use of
> "if(Literal.getLanguage() != null)" in any current code, and if a
> language tag exists, "Literal.getDatatype()" will now return
> rdf:langString so you need to make sure that you check for
> "getLanguage().isPresent()" before checking datatypes. You will be
> able to pick up most of the getLanguage() != null cases by checking
> compiler errors.
>
> The other incompatibility will be that because Literal.getDatatype no
> longer ever returns null, you will need to work on a database
> migration strategy for existing Literal instances that have null, and
> replace them with xsd:string if the language is missing, and
> rdf:langString if the language exists.
>
> URI and IRI are interchangeable in terms of RDF4J.
>
> A fuller migration guide is at:
>
> http://docs.rdf4j.org/migration/
>
> You may want to split up the migration into steps, by stepping up to
> Sesame-2.8 first to pick up the getDatatype() != null cases first,
> then go to Sesame-4 to pick up the getLanguage() != null cases, before
> going to RDF4J with what are virtually all package name changes.
>
> Cheers,
>
> Peter
>
>
> On 29 March 2017 at 12:12, Gustavo Mora <gm...@gmail.com> wrote:
> > Dear,
> >
> > I'm a student of Computer Science, and I'm working on my application to
> > GSOC 2017, but I have some questions. BTW, I also compiled the code and
> > started playing with it. Firstly, there's a plan to migrate because there
> > are some classes that have replacements; for example, IRI instead of URI,
> > changes in some interfaces/classes names, use of optional, etc. Also,
> some
> > tests will fail because RDF4J stores a default string as literal, for
> > example. If we changed that, the next release is not going to have
> > compatibility with the previous ones? If yes, Is it necessary to maintain
> > some deprecated code? What other problems can I encounter in KiWi and
> > platform?
> >
> > Regards,
> > Gustavo Mora.
>

Re: GSOC - Port marmotta RDF4J

Posted by Peter Ansell <an...@gmail.com>.
Hi Gustavo,

One major compatibility issue will be the use of
"if(Literal.getLanguage() != null)" in any current code, and if a
language tag exists, "Literal.getDatatype()" will now return
rdf:langString so you need to make sure that you check for
"getLanguage().isPresent()" before checking datatypes. You will be
able to pick up most of the getLanguage() != null cases by checking
compiler errors.

The other incompatibility will be that because Literal.getDatatype no
longer ever returns null, you will need to work on a database
migration strategy for existing Literal instances that have null, and
replace them with xsd:string if the language is missing, and
rdf:langString if the language exists.

URI and IRI are interchangeable in terms of RDF4J.

A fuller migration guide is at:

http://docs.rdf4j.org/migration/

You may want to split up the migration into steps, by stepping up to
Sesame-2.8 first to pick up the getDatatype() != null cases first,
then go to Sesame-4 to pick up the getLanguage() != null cases, before
going to RDF4J with what are virtually all package name changes.

Cheers,

Peter


On 29 March 2017 at 12:12, Gustavo Mora <gm...@gmail.com> wrote:
> Dear,
>
> I'm a student of Computer Science, and I'm working on my application to
> GSOC 2017, but I have some questions. BTW, I also compiled the code and
> started playing with it. Firstly, there's a plan to migrate because there
> are some classes that have replacements; for example, IRI instead of URI,
> changes in some interfaces/classes names, use of optional, etc. Also, some
> tests will fail because RDF4J stores a default string as literal, for
> example. If we changed that, the next release is not going to have
> compatibility with the previous ones? If yes, Is it necessary to maintain
> some deprecated code? What other problems can I encounter in KiWi and
> platform?
>
> Regards,
> Gustavo Mora.