Contributed tasks aren't added to recent

[alexr00]

This PR fixes #94547

[alexr00]

Background info:
An extension contributed task gets a recently used key that notes that it is contributed.
However, now when recently used tasks are displayed we show them in the same way that tasks.json tasks (custom tasks) are shown (because that's the way we store them as recently used, just like in tasks.json).
This means that the key that the task is saved with doesn't match the key that we try to access it with later.

The fix is to save the task as recently used with the correct custom key instead of the incorrect contributed key.

[dbaeumer]

Why is the key a random key from the customized hash table ?

[alexr00]

The key is from here

vscode/src/vs/workbench/contrib/tasks/common/tasks.ts

Line 731 in 8918d05

public getRecentlyUsedKey(): string | undefined {

I start with a ContributedTask key, but I'm storing a CustomTask, so when I re-create the task later it will have a CustomTask key. To avoid this, I convert the ContributedTask into a CustomTask then use that key.

[dbaeumer]

What I actually mean is the following: how can this piece of code be deterministic given that we loop over the keys in a hash table. There is no order associated with this. Or does it simply not matter.

[alexr00]

There will be only one key in the hash table, since only one task is fed in. I will make that clearer.

[dbaeumer]

OK. That explains it to me. Thanks.

[dbaeumer]

What happens now if we do have a programming error somewhere and the keys are not the same. The we use the key from the beginning which to my understanding would be wrong. So shouldn't we simply assert this or ignore the task if something goes wrong.

[alexr00]

Yes, it is effectively ignored later when the recents are read. However it would be better to ignore earlier (here) instead.

[dbaeumer]

OK. That makes sense now to me. Are you planning to do any changes or do we go with the ignore later approach?

[alexr00]

I made the change for ignoring here since it is better not to defer the problem.

[alexr00]

Thank you for the review!