Relationships/properties/nodes definition

[pablopareja]

Hi!

I just started to implement the definition of relationships in the program InitBio4jTitan specifying cardinality and uniqueness consistency locks. There's something I'm not clear about which is what to do with properties definition. I'm afraid we should maybe define all of them explicitly?? :cold_sweat: :scream: Since we sort of already decided to deactivate automatic type creation I read it's recommended to do it explicitly. What do you think? Here's the link to the commit I'm referencing: d7faca9

[laughedelic]

One important concern here: all these definitions are done in InitBio4jTitan and I believe that it's wrong, because it's bad for modularity. I suggest to separate this file on the parts that correspond to Bio4j modules, so that particular properties/relations will be created only when they are needed. These parts of init can be separate (corresponding to each import program) or just added to the beginning of the import programs (which may be not so good for big parts like uniprot). I think, it will be also easier to maintain it this way. And of course it will simplify some things for the statika-bundles project. What do you think, @pablopareja?

[pablopareja]

mmm I'm not so sure about this being a good idea.... Besides, different modules partially share the same initialization requirements. @eparejatobes WDYT?

[laughedelic]

what is the shared part?

[eparejatobes]

We should separate all type definitions and module-specific init code into module-specific classes. And this should of course match bio4j/bio4j-model#5

We (@laughedelic and me) discussed it past Friday; @laughedelic prepare a short (10 min) talk about how this can be done and where it'd go (API modules etc) and we can see this tomorrow morning OK?

[laughedelic]

ok

[pablopareja]

OK, actually in the case it's done right it can be a pretty cool thing. Nice to see how all this is getting more professional every day! :smiley:

[laughedelic]

See bio4j/bio4j-modules#24

[pablopareja]

What do you want me to see there?

[laughedelic]

I just prepared a talk, as Edu said. For the meeting this morning. But we have problems with the internet in the office..

[pablopareja]

So then?

[laughedelic]

Tomorrow?

[laughedelic]

One suggestion here:

• if possible, all these type definitions should be idempotent, i.e. shouldn't do anything if the key is already created.

[eparejatobes]

agreed

[pablopareja]

OK :+1:

[eparejatobes]

closing