> From: Tom M0LTE via wsjt-devel [mailto:wsjt-devel@lists.sourceforge.net] > Sent: lauantai 13. toukokuuta 2023 10.58 >While manual testing could probably never be eliminated in the case of hamlib, >has there been any consideration of creating test harnesses to remove/reduce >the need for manual testing?
>It strikes me that there could be potential to capture test cases from manual >testing with real rigs, then at least there would be regression tests. Future >bug fixes could have test cases added to prevent further regression, again >without requiring physical radios and manual testing. Hi Tom, Your proposal could work in an ideal world. There are just minor requirements before it is easily done. #1 There should be a (well) defined and agreed CAT protocol that all rig manufactures follow. #2 All rigs should have tested against that protocol. #3 Resources to design and prepare the test program. #4 Resources to design and prepare the Hamlib testing harness against the agreed CAT protocol. On #1 we have a subset of commands that are commonly used. Even no agreed minimum responses to commands are agreed. There is no performance requirements such as timing. The #2 is a farfetched dream. The #3, hups, did I mentioned this? The #4, just what! In addition there are in some implementations bugs and the public CAT command set behaves differently than assumed. All those deal to a situation that the manual testing of most radios is the best way to perform this testing task. For that we need is the famous 'somebody' to keep it going and reporting to the Hamlib task force. Sorry of being to pessimistic, I have some experience on that kind of issue and now safely retired. 73, Reino OH3mA _______________________________________________ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
