This is a fine approach when writing new software or new pieces of software in an existing codebase.
It is not universal though: for a piece of software like KiCad, it is often a significant challenge to come up with a suitable test that actually tests for desired behavior. It is also not currently feasible to write automated tests for parts of KiCad (for example, anything involving the UI). A test that tests against the behavior you *think* the code should have is only useful if that actually matches the behavior the code *actually* should have. For large codebases like KiCad that have been around for a while, understanding the latter takes time working with the code. -Jon On Mon, Jul 14, 2025 at 5:51 AM Salvador E. Tropea <[email protected]> wrote: > Hi Jon: > > My 2 cents on this: > > On 13/7/25 18:03, Jon Evans wrote: > > ... they will easily invent solutions that do not work or have subtle > > problems. > > > I strongly advice to ask to the AI to generate code to test the > generated code. AI can write code really fast, so generating the tests > is fast, and you can be more confident about what was generated looking > at the test. > > > Regards, SET > > -- > You received this message because you are subscribed to the Google Groups > "KiCad Developers" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion visit > https://groups.google.com/a/kicad.org/d/msgid/devlist/1ddb19ac-ce7a-45b9-9de9-075396048d4b%40inti.gob.ar > . > -- You received this message because you are subscribed to the Google Groups "KiCad Developers" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/a/kicad.org/d/msgid/devlist/CA%2BqGbCDpkK0fT8JZNE1NkfWmKpCZWKP-jQFPKuGb8D45JBdP0A%40mail.gmail.com.
