Re: [Patch]Add tab completion for DELETE ... USING
| От | Tatsuya Kawata |
|---|---|
| Тема | Re: [Patch]Add tab completion for DELETE ... USING |
| Дата | |
| Msg-id | CAHza6qdChnuqT15axbOy6jH6E7WHixisNebmFdN=rU38riVKNA@mail.gmail.com обсуждение исходный текст |
| Ответ на | Re: [Patch]Add tab completion for DELETE ... USING (Kirill Reshke <reshkekirill@gmail.com>) |
| Список | pgsql-hackers |
Hi!
Thank you for your review!
> HI! What makes you think we need additional tab completion tests at all?
> If we consult src/bin/psql changes history [0], it can be easily seen
> that not every change to tab-complete.in.c is made with a new test.
> This is (my own understanding, I may be wrong) because we only need to
> test tab-complete patterns that are functionally different, not mirror
> all cases from tab-complete.in.c in the TAP test.
I agree that not every tab-complete change requires tests, and we don't need to mirror all cases from tab-complete.in.c. I also think tests that provide little benefit compared to their maintenance cost can be omitted. However, since this patch implements a previously unimplemented feature (removing a XXX TODO comment), I believe the primary test cases should be kept to verify the core functionality and prevent regressions.
That said, I can reduce the test cases. The current 4 tests could be reduced to 2:
One for USING <TAB> (core functionality: table completion)
One for USING <table> <TAB> (keyword completion after table)
WDYT?
Regards,
Tatsuya Kawata
Thank you for your review!
> HI! What makes you think we need additional tab completion tests at all?
> If we consult src/bin/psql changes history [0], it can be easily seen
> that not every change to tab-complete.in.c is made with a new test.
> This is (my own understanding, I may be wrong) because we only need to
> test tab-complete patterns that are functionally different, not mirror
> all cases from tab-complete.in.c in the TAP test.
I agree that not every tab-complete change requires tests, and we don't need to mirror all cases from tab-complete.in.c. I also think tests that provide little benefit compared to their maintenance cost can be omitted. However, since this patch implements a previously unimplemented feature (removing a XXX TODO comment), I believe the primary test cases should be kept to verify the core functionality and prevent regressions.
That said, I can reduce the test cases. The current 4 tests could be reduced to 2:
One for USING <TAB> (core functionality: table completion)
One for USING <table> <TAB> (keyword completion after table)
WDYT?
Regards,
Tatsuya Kawata
В списке pgsql-hackers по дате отправления: