Key metadata improvements #14022
Key metadata improvements #14022
Conversation
…like tuning ore advanced modes we cannot parse.
|
Rebased to resolve conflicts. |
| enum class [[nodiscard]] UpdateResult{ | ||
| /// The value has been updated and changed. | ||
| Updated, | ||
| enum class UpdateResult { | ||
| /// The value has been updated and changed. | ||
| Updated, | ||
|
|
There was a problem hiding this comment.
This tells me that the enum was deliberately made for error handling. Specifically Rejected should be checked because it almost always would require handling from the caller. This removes that feature, making the error checking optional (which I hope we can agree is very bad practice).
I need more explanation why its appropriate to weaken the guarantees this enum was supposed to provide.
There was a problem hiding this comment.
Concretely: What is your rationale for removing the checks you removed in this PR?
There was a problem hiding this comment.
We have discussed this earlier and conclude that [[nodiscard]] is usefull if discarding is an error. In this case it is only an enum. Discarding can't be an error. It might be an error depending on the returning function. In our case it is updateGlobalKeyNormalizeText() wich works well without error checking. So the change allows us to remove static_cast<void>() on these calls. See at dlgtrackinfomulti.cpp
The counter example are places like this where [[nodiscard]] heps to avois programming errors:
mixxx/src/util/compatibility/qmutex.h
Line 32 in a4e61f9
There was a problem hiding this comment.
In this case it is only an enum. Discarding can't be an error.
Is it though? It seems that Rejected and Unchanged did need some special handling in some cases (which you have removed). I'm asking specifically why its fine to remove the handling for those cases here.
There was a problem hiding this comment.
I think so. I have left an inline comment:
#14022 (comment)
| m_trackRecord.updateGlobalKeyNormalizeText( | ||
| keyText, | ||
| mixxx::track::io::key::USER); | ||
| displayKeyText(); |
There was a problem hiding this comment.
Here we need to call displayKeyText() independent of the result of updateGlobalKeyNormalizeText() because we either want to see the normalized value or the old.
These are the leftover commits form the key metadata handling PRs for #10890
They do not fix anything, but make the code more consistent.