[GitHub] [trafficcontrol] rob05c edited a comment on issue #4291: Added documentation for atstccfg

[GitBox <gi...@apache.org>]

rob05c edited a comment on issue #4291: Added documentation for atstccfg
URL: https://github.com/apache/trafficcontrol/pull/4291#issuecomment-585003768
 
 
   >we should be striving to make testing ATS config gen as easy as possible
   
   That's exactly my point: it _looks_ easy, but it's not _correct_. It's deceptive and insidious (like so much Perl). ORT and atstccfg are both complex, and have complex interactions. For example, ORT has diff checks, which may correctly call atstccfg, but then never place the file if it incorrectly thinks there wasn't a diff. Another good example is the recent caching bug: a single `atstccfg` run wouldn't catch it, and in fact any number of combinations of `atstccfg` file runs would have no issue (I'm speaking from experience – I can't begin to express how maddening it was to track that down). Only if a person (or Integration Test) ran a full ORT would the issue show up. Or ORT may pass unexpected parameters, or not call `atstccfg` at all, or manipulate its output, or any number of other things.
   
   >using atstccfg directly should be made as easy as possible
   
   I still maintain that no person should ever run `atstccfg` directly. I get that we don't have an Integration Test framework for ORT yet, and besides smaller scoped tests are so much easier they're sometimes worth the inaccuracy. But I still maintain that unless they're developing it, nobody should ever run it directly. It _looks_ easy, but it's frequently deceptively wrong.

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services