Hello @eparejatobes @laughedelic @evdokim. Sorry for not giving any sign of living but I was extremely busy(Master thesis, in 2 weeks I am mooving to switzerland etc). Now I am going for week on holiday but after it I will back to work on dynamograph. One more time sorry for it. just a little more patience.
hi @alberskib! looks like things are quite busy on our side as well ") we were working hard on pointless, tabula and scarph, but on the way we got new ideas and have new plans on improving things. quoting yourself
just a little more patience
so don't worry, take your time :wink:
btw, how's your master thesis?
and yes, wish you great holidays!
Hello @laughedelic
I come back from holiday and now I am in Switzerland. I will start my contribution to project. If you have any clues, notes, message (sth like this) do not hesitate to send me. I will start to looking into your changes,
When it comes to the master thesis I am still working on it. Basically all experiments are finished but I need to spend some time to write text, but you know: person cannot just take a seat and spend several hours on writing text - you need a break for writing some code :wink:
I come back from holiday and now I am in Switzerland. I will start my contribution to project. If you have any clues, notes, message (sth like this) do not hesitate to send me. I will start to looking into your changes,
When it comes to the master thesis I am still working on it. Basically all experiments are finished but I need to spend some time to write text, but you know: person cannot just take a seat and spend several hours on writing text - you need a break for writing some code :wink:
Hello @eparejatobes @evdokim @laughedelic. I take a look into cosas (former pointless) and it perfectly fit for dynamograph - It is ready for that? What is more I see that you launch some new projects that seems to be interesting i.e angulillos, I noticed that scarph becoming more and more powerful :smile:
Hi @alberskib! We were away with @evdokim for a couple of weeks (attending a workshop in London), so not much things have changed since then.. :smile:
Now we are back and going to work on all this.. I’m trying currently to bring scarph to the "first-release" state. Also tabula was planed to be reworked (again) and since the decision it was not touched (other priorities). cosas (btw, what do you think about the name?) is in quite a nice state at the moment — a lot of effort was put to split all the abstract parts of other projects into this one and make it consistent..
And about angulillos, I think only @eparejatobes can tell something meaningful.
Now we are back and going to work on all this.. I’m trying currently to bring scarph to the "first-release" state. Also tabula was planed to be reworked (again) and since the decision it was not touched (other priorities). cosas (btw, what do you think about the name?) is in quite a nice state at the moment — a lot of effort was put to split all the abstract parts of other projects into this one and make it consistent..
And about angulillos, I think only @eparejatobes can tell something meaningful.
Hi! Yes, sure, we’re working on moving everything to cosas. If I remember right, scarph
About tabula, I don’t know, of course we will move it to cosas, but other plans... I think we should discuss it with @eparejatobes.
0.5.0-SNAPSHOT is on cosas already. Maybe I should release that version, it worked I think. I’m just working already on ohnosequences/scarph#22 and forgot about that version..About tabula, I don’t know, of course we will move it to cosas, but other plans... I think we should discuss it with @eparejatobes.
About dynamograph I think we should wait for scarph and tabula, I don't want to have yet another migration etc :)
@alberskib about angulillos is something similar to scarph in some sense; it is (I think) close to the limit of what you can do in Java. We are polishing a new Java version of the main Bio4j distribution based on it. It would be nice to implement a DynamoDB backend for it. If you are interested we could think a bit about it, it should be pretty easy; we could also use it as an idea for GSoC15
maybe I could give a presentation on angulillos
Hey @eparejatobes. DynamoDb backend for java version is also interesing for me - I'm in. I almost finish first draft of my master thesis so in couple of days I will much more time. Regarding this presentation on anulillos - there will be some streaming or slides? I would like to watch it but probably my time availability is not sufficient (evenings) so I could take a look into presenation or recording.
@alberskib I was thinking that it'd be a good idea if you could take a look at the current state of both https://github.com/bio4j/angulillos and https://github.com/bio4j/angulillos-titan, and ask me if there's something that you don't understand there. Overall the library is pretty straightforward if you factor out the java generics noise :)
I think it should be pretty straightforward to implement bio4j/angulillos-dynamodb, and it would let us experiment with DynamoDB as a backend with real big data. The scarph stuff is really nice but I honestly cannot give you an estimate of when it will be ready; so I want to keep things going regarding DynamoDB and your project.
Hello @eparejatobes. Today i just took a quick look and at the begining I had one question. I was diving into code and after some time I think that I found answer for my question but maybe it will be better if you answer it - just to be sure: Could you tell me why i.e. TypedEdge have parameter SourceGraph (SG)/TargetGraph (TG) ? In a weekend I spend some more time.
It’s better if @eparejatobes answer this, but I have a quick guess :wink:
I think it’s because it was developed with bio4j in mind, so bio4j has different parts/modules/subgraphs (like GO, Uniprot, etc.) and we want to have each such graph independent, while being able to connect them — so an edge can go from a (source-)type in one graph to a (target-)type in another graph.
I think it’s because it was developed with bio4j in mind, so bio4j has different parts/modules/subgraphs (like GO, Uniprot, etc.) and we want to have each such graph independent, while being able to connect them — so an edge can go from a (source-)type in one graph to a (target-)type in another graph.
At first I'd be more interested in doing the writing part; reading is easy
@/all Hey guys!
First of all I would like to tell you that I am feeling terrible that I was constantly saying that I am starting but I did not produce any progress. Sorry for that!
Today I come back and I will do everything(I have job so I will be working mostly on evenings and weekends) to deliver first version as soon as possible. From now I am starting work on that (regularly) to deliver systematicly product increments. I think that at some point I will need little help from you.
One more time sorry for that.
First of all I would like to tell you that I am feeling terrible that I was constantly saying that I am starting but I did not produce any progress. Sorry for that!
Today I come back and I will do everything(I have job so I will be working mostly on evenings and weekends) to deliver first version as soon as possible. From now I am starting work on that (regularly) to deliver systematicly product increments. I think that at some point I will need little help from you.
One more time sorry for that.
Hey. I think that I am missing part of big picture - as a result I have next questions:
- Do I need to override
TypedEdge,TypedVertexin order to provide specific versions for DynamoDb backend or just implement some kind ofDynamoDbRawVertex,DynamoDbRawEdge. I am not sure: I think that yes I need to, but when I look into titan or neo4j implementation of angulillos there is no such files. IinTypedEdgeandTypedVertexis just for identifying used backend and offer possibility to communicate graphs from different backends?
gitter notifications are pretty buggy
First, all the actual interaction happens at the UntypedGraph level
You do need to define classes like
DynamoDBRawVertex, DynamoDBRawVertexType, etc
As here we're free to implement all those things as we please, we can take a more reasonable approach.
Tables means just table names, of course
All the
XXXType classes should be completely decoupled from any resource management, auth etc.
That should be at the untyped graph level
About raw edges and vertices, they should have just a link to the corresponding type together with something identifying them (the property id used in the corresponding table , for example)