2018-10-01 Meeting notes

Tim Whitlock (Deactivated)

• QBL Blocks. Fran has a list of fixes that this release would also include for checking. foghorn_leghorn (Unlicensed) wil liase with her over testing and releasing this code. Currently awaiting Frans response.
• EPO dosage data has been imported and the data base definitions updated for the fileds. (Int to Float). This means current Validation code will not work until updated. with the EPO fix. This could be done as part of the QBL block release or as a separate interim fix, foghorn_leghorn (Unlicensed) will check with Fran her preference and make sure she is aware that the validation code will not work correctly until the fix is applied.
• Kings pdf issue. George Swinnerton has found the issue seems to be caused because the current pv-rda code re-encodes the base64 encoded pdf data again resulting in hte rda file holding the pdf base64 encoded twice. This results in the data being stored as base64 so the pv-extract code assumes the data is already encoded. The kings rda file is only base64 encoded once (as would be expected meaning the data would then be stored as binary in the database as a result pv-extract needs to base64 encode it before sending it. This is an error in the pv-rda code that needs correcting. Once corrected all the letters need to be re-saved as binary and not base64 encoded. George Swinnerton will make these changes and test. Once the issue with pdfs is sorted it should be possible to sort other issues with the xml. One noted so far is there is a gp element even when no data is present. This fails validation.
• Extract Specification has not been completed and will be this week including the list of rda fields which conflict with the generic PV testnames which encompass several RDA fileds which would be required in their RDA separate entity entries for the RR Extract to work.
• George Swinnerton mentioned Bristol had identified an issue with letters on the same day with no time element not being stored separately. The currrent process does not take into account letter content so two letters on the same day with no differing time element will be treated as the same letter and one will overwrite the other.This is a repository issue and a ticket has been raised for Nick https://jira.ukrdc.org/browse/TNG-483?jql=text%20~%20%22letters%22.
• rapolas (Unlicensed) raised the question of Gender displaying as unknown in RaDaR due to it defaulting to the latest result from PV in the lists even though there may be a Gender correctly entered in RaDaR and it is just the PV feed that has an issue (mainly Cardiff who don't seem to send the gender field). The impact of this is seen in the INS and CKD queries which report unknowns even though the RaDaR record is populated. George Swinnerton will check his code is correctly using the data models to ensure Gender is populated if present. It was suggested that the UI should not display Unknown if a Gender was available in the RaDaR record but not in the lates PV feed. rapolas (Unlicensed) look at modifying how the field is shown so that it will show a value if one is found not just the last received.

 George Swinnerton reported Anna wanted the data today. George Swinnerton was not sure if any data quality checks would be done/fixed by the Data managers prior to the release to statistics.

An IBD url referenced by a number of tests in PV has been allowed to expire. This is because the server was set up by a phd student who has moved on and no one then renewed the domain according to Nadim. Until a replacement is in place George Swinnerton has removed the links.

George Swinnerton reported the search option on github has been restored after it broke when the data breach was fixed by github.