1. 12 Jul, 2019 1 commit
  2. 11 Jul, 2019 4 commits
  3. 10 Jul, 2019 4 commits
    • colemanw's avatar
      78f3f7c0
    • colemanw's avatar
      224b94cc
    • totten's avatar
      GenCode, Cache::cleanKey() - Fix deploop during clean initialization · 77f080cb
      totten authored
      Overview
      --------
      
      Suppose you have a clean codebase with no DAO files, and you try to run `xml/GenCode.php`.
      This currently crashes.
      
      The problem is not symptomatic day-to-day because we commit DAOs, so it's
      very rare to run a full/clean initialization.  However, it does get in the
      way of stress-testing the correctness of GenCode.
      
      Before
      ------
      
      Since 76e697a9, I believe we've had a dependency-loop
      in the clean-gencode scenario, which works as follows:
      
      * We don't have any DAOs, so we run `GenCode`.
      * Running `GenCode` does a partial bootstrap (i.e. `CRM_Core_Config::singleton($loadFromDB===FALSE)`)
      * The partial bootstrap creates some thread-local caches (`Arraycache`)
      * Before creating the cache, it normalizes the name by calling `CRM_Core_BAO_Cache::cleanKey()`
      * `CRM_Core_BAO_Cache` extends `CRM_Core_DAO_Cache`
      * `CRM_Core_BAO_Cache` doesn't exist - that's why we called `GenCode` at the beginning.
      
      After
      -----
      
      This migrates the utility function `cleanKey()` to another class which is always available and
      does not rely on DAOs.
      77f080cb
    • eileen's avatar
      [REF] simplify & add tests on getMappingParams · d67808ac
      eileen authored
      Tests extended to cover search builder (all that operator stuff
      d67808ac
  4. 09 Jul, 2019 5 commits
  5. 08 Jul, 2019 3 commits
  6. 05 Jul, 2019 3 commits
  7. 04 Jul, 2019 3 commits
  8. 03 Jul, 2019 2 commits
  9. 02 Jul, 2019 1 commit
    • eileen's avatar
      [REF] Cleanup fixSchemaDifferencesFor() · fcc20cec
      eileen authored
      This is always called with FALSE as the last param except from the unit test - the test can do it's own work on it
      Also passing NULL for param 2 is the same as not passing anything
      fcc20cec
  10. 01 Jul, 2019 7 commits
    • eileen's avatar
      [REF] remove instances of pass-by-reference where no change takes place · 6477f387
      eileen authored
      I did a grep on these and the passed params are not altered - down with php4
      6477f387
    • eileen's avatar
      Consolidate handling of conflicts between the batch job and get_conflicts api · ffa59d18
      eileen authored
      This ensures that conflicts are stored during batch_merge to the prev_next cache with the same format as when the api
      calls get_conflicts. The code doing this wrangling is moved from the api to the BAO layer.
      
      We add a test to ensure the output is the same & use the previously added test to check the string is the same too.
      
      Test cover here is pretty good
      ffa59d18
    • eileen's avatar
      [REF] move formatting of conflict into markConflict function · 5214f03b
      eileen authored
      The goal here is to make this information more usable to an api caller. Currently the 'conflict' key holds text formatted
      for a specific form. I was going to change this to hold an array of data that an api could use. However, the
      Contact.get_merge_conflicts already has a specific api-friendly format that is keyed by 'mode'
      at the top level so my revised plan is to return api-friendly data as well (without messign with the
      existing form's data).
      
      This change simply moves the decisions about what data to store & how down to the function whose responsibility
      it is (markConflict) and adds a test to ensure that no change results
      
      [REF] move formatting of conflict into markConflict function
      
      The goal here is to make this information more usable to an api caller. Currently the 'conflict' key holds text formatted
      for a specific form. I was going to change this to hold an array of data that an api could use. However, the
      Contact.get_merge_conflicts already has a specific api-friendly format that is keyed by 'mode'
      at the top level so my revised plan is to return api-friendly data as well (without messign with the
      existing form's data).
      
      This change simply moves the decisions about what data to store & how down to the function whose responsibility
      it is (markConflict) and adds a test to ensure that no change results
      5214f03b
    • Andrew Hunt's avatar
      4a413eb6
    • eileen's avatar
      [REF] extract prepareCreate from CustomField.create · 4f166aec
      eileen authored
      Straight forward extraction - cleaning up code with a view to offering a more efficient wrapping function
      4f166aec
    • mattwire's avatar
    • eileen's avatar
      Block classes in unserialize field for IDE cheer · 8cec96dc
      eileen authored
      8cec96dc
  11. 29 Jun, 2019 1 commit
  12. 28 Jun, 2019 1 commit
  13. 27 Jun, 2019 5 commits