Prior Analyses are displayed yet derived results unreachable when Dataset version is incremented

[scottx611x]

• Commit: b301acb
• Locally & on AWS

I'm not sure if what I'm describing below is by design hence thebug & question labels

---

Steps to reproduce

1. Upload a Dataset. I used hg19-metadata-s3.txt
2. Run any Analysis successfully on said Dataset. You should see your Analysis in the Analyses listing tab as shown below:
   
3. One should also be able to see the Derived Data files from this Analysis when browsing to the Files tab.
4. Now re-upload hg19-metadata-s3.txt once more under the same username.
5. Ensure in the Details tab that the Dataset's version has been incremented from version 1 to version 2
6. Run another successful Analysis like before.
7. This time you should see both Analyses in the Analyses listing as shown below
   

Observed behavior

• Although all prior Analyses from version 1 of the Dataset are listed under the Analyses tab, one is only be able to see the Derived Data files from the 2nd Analysis on version 2 of the Dataset when inside of the File Browser.

Expected behavior

• Not sure. It seems odd that one wouldn't be able to see the results of prior Analyses, but maybe this is by design.

NOTE:

• I find the Dataset versioning to be confusing and end-users may be confused as well.
• Maybe we could support viewing older versions of datasets?

[scottx611x]

I just ran into this again. Its particularly annoying because the derived results exist, but aren't accessible.

A scenario I could imagine where this could commonly occur:
Within a Group: A & B:

User: A uploads Dataset: A and Runs a successful Analysis on it. User: A is happy with the derived results.

Some time passes by and User: B re-uploads the same metadata for DataSet: A incrementing its version (most likely without the knowledge of doing so) and runs a successful Analysis on it. User: B is happy with the derived results not knowing that they have just made User: A's derived results inaccessible to the entire Group: A & B.

[jkmarx]

Dev meeting discussions: @jkmarx Write issue: There's back end support for rolling back versions, just need a UI interface. @scottx611x Double check the behavior for running analyses on data sets (cross contamination) Check with modify perms: upload same meta data and see if it overwrites.