CFE Inline Integrity Checks: Difference between revisions

From CCMDB Wiki
Jump to navigation Jump to search
mNo edit summary
 
(16 intermediate revisions by 4 users not shown)
Line 1: Line 1:
===CFE editing - need integrity checks in place to catch editing errors===
Pagasa edits in CFE, and right now there are no checks in place for typos in these edits.
*Admit transfer discharge range as from CCMDB


=== Why not to do this inline? ===
== When and how would these checks run? ==
* What would trigger the checks? We don't have "complete" to trigger a check, so what would trigger it? on_change has problems when used on comparison of multiple fields.
Cross-checks in CCMDB are triggered by collectors checking the various "complete" boxes, but there is no equivalent to this in CFE, especially since Pagasa can legitimately edit tables directly or via query, so these checks would likely need to be triggered manually, likely between [[Quality Assurance queries in CFE]] and actually setting records to "vetted".
'''Best to not do this inline after all.'''
 
* We'll want to add some "post processing checks" that Pagasa would run after all edits for a day. To be done between [[Quality Assurance queries in CFE]] and actually setting records to "vetted".  
== What sort of checks do we actually want? ==
Pagasa agreed that the most likely items to be changed that need to trigger a check are [[Service tmp entry]] and [[Boarding Loc]] and [[Transfer Ready DtTm tmp entry]] and [[Dispo DtTm]]s, and they need to be checked.
 
* Possible checks:
** All S, B, T have to be before D
** any T has to be after the B it's linked to
*** how about T the same as B it's linked to?  see cases like this. should we allow? or T must increment one minute from B.  --[[User:JMojica|JMojica]] 10:11, 2022 August 10 (CDT)
** any S/B has to be before any CCI and ICD10, and D has to be after these
* Pagasa, Julie, can you think of other things to check?
** First S must be the same as first B.  - --[[User:JMojica|JMojica]] 10:11, 2022 August 10 (CDT)
** Any CCI and ICD10 with datetime must have a corresponding B date.  - --[[User:JMojica|JMojica]] 10:11, 2022 August 10 (CDT)
** do we allow first B ER the same as second B non-ER when previous loc is ER?  or simply delete the first B ER from tmp if the same as second B non-ER. - --[[User:JMojica|JMojica]] 10:11, 2022 August 10 (CDT)
** only one T per B it's linked to (no multi T).  each T must have an integer.  found case with mutli T- one without integer and another with integer. - --[[User:JMojica|JMojica]] 10:11, 2022 August 10 (CDT)
** T should not have missing time or should not have check mark.  - --[[User:JMojica|JMojica]] 10:11, 2022 August 10 (CDT)
 
The checks could run at change time, or just before vetting.


===What sort of checks do we actually want? ===
{{Todo  
{{Todo  
  | who = Tina  
  | who = Tina  
  | todo_added = 2021-07-27
  | todo_added = 2022-09-06
  | todo_action = 2022-08-09
  | todo_action = 2022-09-06
  | question =
  | question = _dev_CFE
* * This would best be based on things Pagasa is likely to check and change
* ready to implement inline cross check for CFE edits. [[User:Ttenbergen|Ttenbergen]] 10:03, 2022 September 6 (CDT) }}
** [[Service tmp entry]] and [[Boarding Loc]] and [[Transfer Ready DtTm tmp entry]] and [[Dispo DtTm]]s that are not chronological are most likely culprits.
** All S, B, T have to be before D
** any T has to be after its B
** All CCI and ICD10 have to be after first S/B and before Dispo
}}


== Related articles ==  
== Related articles ==  

Latest revision as of 10:03, 2022 September 6

Pagasa edits in CFE, and right now there are no checks in place for typos in these edits.

When and how would these checks run?

Cross-checks in CCMDB are triggered by collectors checking the various "complete" boxes, but there is no equivalent to this in CFE, especially since Pagasa can legitimately edit tables directly or via query, so these checks would likely need to be triggered manually, likely between Quality Assurance queries in CFE and actually setting records to "vetted".

What sort of checks do we actually want?

Pagasa agreed that the most likely items to be changed that need to trigger a check are Service tmp entry and Boarding Loc and Transfer Ready DtTm tmp entry and Dispo DtTms, and they need to be checked.

  • Possible checks:
    • All S, B, T have to be before D
    • any T has to be after the B it's linked to
      • how about T the same as B it's linked to? see cases like this. should we allow? or T must increment one minute from B. --JMojica 10:11, 2022 August 10 (CDT)
    • any S/B has to be before any CCI and ICD10, and D has to be after these
  • Pagasa, Julie, can you think of other things to check?
    • First S must be the same as first B. - --JMojica 10:11, 2022 August 10 (CDT)
    • Any CCI and ICD10 with datetime must have a corresponding B date. - --JMojica 10:11, 2022 August 10 (CDT)
    • do we allow first B ER the same as second B non-ER when previous loc is ER? or simply delete the first B ER from tmp if the same as second B non-ER. - --JMojica 10:11, 2022 August 10 (CDT)
    • only one T per B it's linked to (no multi T). each T must have an integer. found case with mutli T- one without integer and another with integer. - --JMojica 10:11, 2022 August 10 (CDT)
    • T should not have missing time or should not have check mark. - --JMojica 10:11, 2022 August 10 (CDT)

The checks could run at change time, or just before vetting.

_dev_CFE

  • ready to implement inline cross check for CFE edits. Ttenbergen 10:03, 2022 September 6 (CDT)
  • added: 2022-09-06
  • action: 2022-09-06
  • Cargo


  • Categories

Related articles

Related articles: