Closed rimas-kudelis closed 2 years ago
BTW I also merged your 2.7 branch back into master.
Also, 52cf95a might be worth cherry-picking into 2.7 as well, since you broke symlinks you had committed previously. Although I don't think you should even commit them: they seem to be created automatically during composer install
.
The PR was closed because I deleted my fork, but I think you can still reopen it if you feel like it.
This (hopefully) fixes #9, albeit with some caveats. Quoting the README, here are the missing features:
pdo-firebird
extension doesn't support this natively. You implemented it by using a query, but it requires the name of a sequence as an argument. Not sure if implementing it in PDO this way would be much of a win over not implementing it at all. Futhermore, I'm not sure if it works within a transaction (I think it was failing for me with a "transaction is already active" or similar message, but that was before I actually fixed transactions, so it's probably not an issue anymore.ext-pdo-firebird
only supports dialects starting with PHP version 7.4, so this driver will also only support it starting on PHP 7.4 and up. Futhermore, only dialects 1 and 3 are supported according to PHP documentation, so even though the test for dialect 2 passes, I'm not sure it always will. And dialect 0? I guess there's no such thing. The test for dialect 0, which expected it to be treated as 3, failed, so I assume in PDO it falls back to 1 instead of 3. Either way, I guess it shouldn't be used or tested against.I also removed some assertions about internal class architecture from the tests. Swapping the backend ruined those assertions. I think we're supposed to test how the unit works, as opposed to how it is constructed.