The permissions module allows for filtering of data from graphs based on an external access pattern. Basically, it uses dynamic proxies to wrap the instance of Model or Graph (and associated beasts like RDFList) so that users can not see data they are not allowed to see. When I wrote that comment I thought there were classes for the Inference Graph and Models but I was mistaken. It simply uses the underlying secured model or graph to execute. So no bother there.
Claude On Thu, Oct 24, 2019 at 12:10 PM ssz <sss.z...@gmail.com> wrote: > The '*.ontapi.jena.UnionGraph' is an analogue of > 'org.apache.jena.graph.compose.MultiUnion', not > 'org.apache.jena.graph.compose.Union', just like 'OntGraphModel' is an > analogue of 'OntModel'. > There are differences between these two graphs: > - structure. 'UnionGraph' requires one base, while 'MultiUnion' does not > - it can be distinct or not (now it is always distinct, but for future > release I'am planning to turn off this option, but only in OWL-API view - > for OntGraphModel it will remain true). > - it is safe for recursions: it can contain itself as a child, while > 'MultiUnion' cannot (i.e. StackOverflowError). > - it is used in other parts of ONT-API (not only in jena subsystem - i.e. > not only in the package 'com.github.owlcs.ontapi.jena') > - it has it is own 'GraphEventManager' with extended functionality > - the 'UnionGraph' is a part of ONT-API, and I can controll it, what cannot > be said about 'MultiUnion' - perhaps, this is the main reason. > > There could be other reasons - I can't recall right now. > It is also possible that there were good reasons that led to the appearance > of this graph, but now these reasons have disappeared. > Anyway, if the investigation will prof that some ONT-API thing can be > safely replaced with Jena-analogue, then such replacement will be > performed. > Migration of 'OntGraphModel' is an example of such possible replacement. > And right now I don't see other replacement candidates. > > I don't know what is permissions layer in Jena; in ONT-API it is unneeded, > but if it is required by Jena, it can be added. > > 'OntGraphModel' can be supported by me, whether it remains as part of > ONT-API, or migrates to Jena, depending on ONT-API needs and bug-reports. > And, of course, depending on free-time. > I'm still working continuously on ONT-API, and so far I am not going to > abandon it. > Again, it is not an urgent question, > some things can be done first in bounds of ONT-API, and migration can be > done later (or not). > > On Wed, Oct 23, 2019 at 11:05 PM Claude Warren <cla...@xenei.com> wrote: > > > OK. I finally had a few minutes to look at this code. > > > > I agree that supporting OWL2 would be nice. > > > > There seems to be a lot here. (I have not read through it all and > > therefore do not have a full understanding of the contribution) I too am > > concerned about having enough developers that understand the code to be > > able to support it. But as more people use it more will achieve the > level > > of competence necessary to support it. I believe that will come with > time > > and as I recall we have had questions about OWL2 support so I suspect > there > > is a population of developers that will contribute and support. > > > > I am concerned about possible code duplication. For example there is a > > UnionGraph class in the contribution, how does this differ from the > > existing Union class? > > > > My other concern is the impact on the permissions layer. This is a new > > Model type and the permissions layer does not directly support it. There > > are additional classes like OntList that should be wrapped. This might > be > > an opportunity to revisit the Permissions layer and determine if there > is a > > better implementation strategy. On the other hand, I don't have the time > > to do a full revisit. I could, however, work with ss zuev to ensure that > > the permissions layer provides support for the new Ontology classes. > > > > I know we normally vote +1, -1 or 0 but in this case I am > > > > +0.5 because I am not certain of the level of support available for the > > code being contributed. > > > > If there is more support from other PMC members I could be convinced to > > change to +1 > > > > On a side topic, do we have a documented procedure for accepting > > contributions? I don't recall what I did for the Permissions layer and I > > know there have been several contributions since then. > > > > Claude > > > > On Wed, Oct 23, 2019 at 7:03 PM ss zuev <sss.z...@gmail.com> wrote: > > > > > >> PMC and community - your thoughts. > > > > Not sure what does it mean, but it doesn't look like important. > > > > > > I apologize, just misunderstood this. > > > Well, If the PMC and community decide that this feature is good to be > > > included as a Jena part, then, I think, the discussion can be moved to > > > Jira. > > > It is not an urgent question. > > > Thank you for your time. > > > > > > > > > On Tue, Oct 22, 2019 at 8:44 PM ss zuev <sss.z...@gmail.com> wrote: > > > > > > > > PMC and community - your thoughts. > > > > > > > > Not sure what does it mean, but it doesn't look like important. > > > > > > > > The new home for ONT-API is the owlcs group, it is discussed [here]( > > > > https://github.com/owlcs/owlapi/issues/889). > > > > The relationship between ONT-API and OWL-API is described in the > > > README.md > > > > of ONT-API. > > > > The question was not about moving the whole project, but only part of > > it > > > > (OntGraphModel), sorry if I didn’t put it clearly enough. > > > > It is the package 'com.github.owlcs.ontapi.jena' (the former > > > > 'ru.avicomp.ontapi.jena') and the main class is 'OntGraphModel'. > > > > All the rest parts of API are definitely not suitable to be moved > into > > > > Jena. > > > > But this one - is a classic Jena Model, small enough, and I thought > it > > > > would be quite convenient for people who are used to the 'OntModel'. > > > > Although it is not 'InfModel', it has rich functionality to work with > > all > > > > OWL2 things, which is what many people want. > > > > In the page https://jena.apache.org/documentation/ontology/ there > are > > > > words about OWL1.1: "We will be addressing this in future versions > > Jena." > > > > so I thought OntGraphModel could be a good start to such support. > > > > > > > > "Copyright (c), The University of Manchester, owl.cs group." is a > > header, > > > > that I set a few days ago. > > > > Instead of old one, since now the project home is owlcs. I can't see > > any > > > > problem here, it seemed to me a pretty good header. > > > > And, of course, I didn't mean the moving everything as it is: this > was > > a > > > > question, while moving the header would be changed to apache license, > > and > > > > classes/packages would be renamed. Sorry if this was unclear. > > > > > > > > > the provenance is unclear. > > > > > > > > Well, the whole provenance is definitely in the git history, no any > > other > > > > places exist. > > > > Maybe it is unclear due to the moving - this operation is not > directly > > > > supported by the git itself, although some tools can glue the commit > > > > history after moving. > > > > > > > > > So an important question is whether, from the wider Jena community, > > > > there are people sufficiently interested to take this forward. > > > > > > > > Well, I can support this model, but in this case, I tend to think it > is > > > > better to leave the 'OntGraphModel' as a part of ONT-API (i.e. within > > its > > > > new home - owlcs). > > > > > > > > So, the question, I think, can be closed. Thank you for your time. > > > > > > > > > > > > On 2019/10/18 10:29:03, ss zuev <sss.z...@gmail.com> wrote: > > > > > Hello, everyone. > > > > > This is the question. > > > > > > > > > > There is a project [ONT-API](https://github.com/avicomp/ont-api), > > > which > > > > is > > > > > a jena-based OWL-API implementation. > > > > > It includes Ontology RDF Model, > > > > > [ru.avicomp.ontapi.jena.model.OntGraphModel]( > > > > > > > > > > > > > > > https://github.com/avicomp/ont-api/blob/master/src/main/java/ru/avicomp/ontapi/jena/model/OntGraphModel.java > > > > ), > > > > > > > > > > that is an analogue of org.apache.jena.ontology.OntModel but for > OWL2 > > > > > semantic. > > > > > > > > > > Currently I can't continue with domain avicomp, and, therefore,I'd > > like > > > > to > > > > > move somewhere the whole project or some its part. > > > > > I think the OntGraphModel as a feature could be very convenient for > > > Jena > > > > > users. > > > > > > > > > > Few additional notes: > > > > > ONT-API has direct tests for only this model (package > > > > > ru.avicomp.ontapi.tests.jena), > > > > > and total number of testcases (which might be considered as > indirect > > > > tests) > > > > > is about 5500 (many of them were taken from OWL-API). > > > > > It also has benchmarks and memmarks (but not for `OntGraphModel`). > > > > > Also it has javadocs and usages: (my) ONT-D2RQ, ONT-MAP projects, > and > > > > > protege-like system (yet not in github), many other users also use > > > > ONT-API > > > > > somehow. > > > > > > > > > > Obviously, the possible moving would mean total refactoring : > > changing > > > > > namespaces and class names, fix docs, maybe moving methods, etc. > > > > > > > > > > What do you think about this? > > > > > > > > > > > > > > > > > > -- > > I like: Like Like - The likeliest place on the web > > <http://like-like.xenei.com> > > LinkedIn: http://www.linkedin.com/in/claudewarren > > > -- I like: Like Like - The likeliest place on the web <http://like-like.xenei.com> LinkedIn: http://www.linkedin.com/in/claudewarren
