D4Science Infrastructure

RUserService has been dismissed, the new name is RConnector

I know nothing about this app, don't even know where we hosting it on our infrastructure and how we auth on it , can you provide me with more details? @lucio.lelii@isti.cnr.it you worked on this if I recall correctly

The current RConnector implementation is TabMan specific. For now it requires a username and a TabularResourceId. What is the use case here ?

I believe the part on connection of a user from D4Science to RStudio should be isolated from the TabMan specific behavior. This would make RStudio reusable without passing for TabMan. In the meanwhile we could use it by specifying a sample dataset that will be loaded by RStudio even if it is not needed.

I have already changed the r-connector server and client. Now the tabular resource ID is not required and the username is taken from the authorization token.
see ticket #2615

@gianpaolo.coro@isti.cnr.it It is unclear to me whether the requirement is to have RStudio embedded into an iFrame within a given VRE or a Page that would open a separate browser tab is acceptable, because what we have now would support the latter option.

http://rstudio-dev.d4science.org

I my opinion, the user would benefit from opening a new browser tab, because otherwise including RStudio in the portal frame would be uncomfortable. The important point is to have the user workspace automatically created by Lucio's service.

I'm releasing the new RConnector in 3.10.1.
here you can find a small documentation : https://gcube.wiki.gcube-system.org/gcube/RConnector

Dear all,
I will try to clarify our needs.

At this stage, we are executing the R codes we want to deploy on the StockAssessment VRE by using an ssh access to the following server : https://dev.d4science.org/group/devvre/statistical-manager . This is a strong limitation for many users of the VRE who would surely like to try as well some execution or customization of (available or new) R codes with data available within the StockAssessment VREs without ssh. At some point, some of these codes brought by users could be made available in the Statman by using SAI. There is no way to execute these codes in the VRE for now before trying a deployment with SAI. Many users are used to RStudio and the SAI doesn't enable to execute the R code on the infra before deploying it but its GUI is very similar to Rstudio. We think RStudio is a missing component to browse and execute R source codes of the VRE and related data.

By having a direct access to RStudio users will find an application they are used to and be able to customize some codes without having to go only through StatMan where parametrization of codes is limited. I understood as well that ICES was using this kind of facility to drive some workshops.

I added Paul as a watcher of this ticket as Paul was already using RStudio (http://rstudio.d4science.org/, from tabman) to test some R codes on the infra. Soon, we would like to test R codes with Sardara database storing tuna datasets. To run such tests in Rstudio Paul has to open a fake dataset (he doesn't care about it) in the TabMan and then only, he can access Rstudio and download the datasets and codes he needs to work with from external URLs (not from the VRE or not from his workspace as we expected). So this is finally very complicated and it sounds that you can finally set it up as a dedicated item in the upper menu of the VRE.

We can discuss this use case if it needs to be clarified. As the Rstudio ling was already in the TabMan, I thought that was already possible.

Dear @julien.barde@ird.fr I understand your point. However, although the Rstudio is already in the TabMan, in order to fulfil your request we needed to release the new RConnector in 3.10.1.. @lucio.lelii@isti.cnr.it completed the activity about the R connector and we are performing tests in the development environment.
The plan is that we are going to add the Studio in stock assessment VRE in the upcoming gCube 3.11.0 Release (before the end of april).

Dear Massimiliano

thanks for your message. This is fine like this, we can wait and we can work in RStudio through Tabman if needed as we know how to do it. This will be helping for newcomers in the VRE but not a blocking issue for our work in the project.

thanks

julien

RStudio is now available and very useful to execute some R scripts.
However, as explained during TCom2, it would be even more useful with additional functionalities:

1. same R configuration (R version and packages) as the one on machine(s) used by Dataminer. This is key for us to validate (without errors) the algorithms we want to deploy by using SAI / dataminers.
2. since SAI / dataminer algorithms will use some R scripts which might be stored on the workspace: could we open some workspace repositories containing R Scripts directly from Rstudio in order to edit directly the files stored in the infra and used by dataminer ?