Changes between Version 48 and Version 49 of Developers/Developing Code Changes


Ignore:
Timestamp:
2019-09-16T14:28:04+02:00 (12 months ago)
Author:
clevy
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Developers/Developing Code Changes

    v48 v49  
    9696* Changes in the reference manual should be also written as soon as possible, and before review process. 
    9797 
    98 ==== Test your Change 
     98==== Test your Changes 
    9999 
    100 Test your change appropriately and record the results. The testing required will depend on the change and the potential impact on other configurations. When testing, things to consider include 
     100Test your change appropriately and record the results in the wiki page for your action. The testing required will depend on the change and the potential impact on other configurations. When testing, things to consider include 
    101101 
    102 * full successful SETTE tests: to check nothing is broken or goes wrong in the reference configurations SETTE tests: if running with some changes in the results, those have to be justified and checked in detail. New code should not change results when it is switched off, and not alter results when tuned on. 
    103 * full successful specific tests: to be defined before starting, in order to check that development does indeed what is expected 
     102* full successful SETTE tests 
     103The SETTE script allow to check in sequence that nothing has been broken for the reference configurations. 
     104This script and the associated information needs to be downloaded: 
     105 
     106{{{ 
     107cd "my working directory" 
     108svn co https://forge.ipsl.jussieu.fr/nemo/svn/utils 
     109cd utils/CI/sette 
     110}}} 
     111* Once you have sucessfuly ran the sette.sh script, the sette_rpt.sh scripts summarizes the results of your test. 
     112 
     113 If you get some changes in results or some errors , those have to be justified and reported in detail in the wiki page for the reviewer. New code should not change results when it is switched off, and not alter results when tuned on. 
     114* full successful specific tests: to be defined during preview phase, including use of associated test case in order to check that development does indeed what is expected 
    104115 
    105116=== Review