Incomplete Type Index / Bug 109084

[HeikoKlare]

Current Behavior

In some cases (with unknown reasons), the type index does not contain all available types, so the type search does not find some types that actually exist.
Rebuilding the index via the preferences does not resolve the issue.

The bug has been existing for a long time and does not seem to be fixed by now: https://bugs.eclipse.org/bugs/show_bug.cgi?id=109084

Expected Behavior

In the best case, the type index should properly contain the expected types.
At least, the "rebuild index" operation in the preferences should properly rebuild the index.

Existing Workaround

Delete .metadata.plugins\org.eclipse.jdt.core*.index, and restart Eclipse.

[fedejeanne]

The problem was reproduced in 2023-03 but it was not because of a broken index so applying the workaround did not solve the problem.
The problem is not reproducible in 2023-06.

[HeikoKlare]

Why is this one closed if the conducted reproduction does not apply to the issue? The original issue was a broken type index (e.g., because of some application crash during JDT indexing). This led to faulty results when using the index (e.g., when using type search). I don't see how a case where no broken index exists shows that this issue is not reproducible or that the workaround is not applicable.

[fedejeanne]

Sorry, I mixed this one with the (not so) related issue: #9

[HeikoKlare]

This is one has been marked as "in work" for half a year: what is the current state?

[fedejeanne]

I left it as "Ongoing" in August because I intended to see if the problem surfaced again, but I see no activity in it and I also have no novel ideas on how to tackle this.
Should we set it as "new" again? BTW what happened to "ready"?

No idea why it wen from Ongoing -> In Progress: short

[HeikoKlare]

No idea why it wen from Ongoing -> In Progress: short

I guess it was moved because it does not fulfill our definition of being "ongoing" :-)

Should we set it as "new" again? BTW what happened to "ready"?

I think it should be moved to "ready" then.

[fedejeanne]

No idea why it wen from Ongoing -> In Progress: short

I guess it was moved because it does not fulfill our definition of being "ongoing" :-)

Do we have our definition of Ongoing written somewhere? It was my understanding that as long as there is something going on but there is no "active" work on the ticket then it's "ongoing". I activated some logging back then to see if there was something useful in the logs when the error happened.

[HeikoKlare]

Do we have our definition of Ongoing written somewhere?

Yes, it's given in the description of the backlog state: