improvement: Retry indexing source file

[tgodzik]

Previously, if there was an issue indeing a file we would fail searching for the particular query. Now, we pack it into try an retry in case of errors such as filesystem ones.

Fixes #5231

[kasiaMarek]

Seems reasonable, though I'm only wondering if the reindexing doesn't happen to fast. Let's say you really have a something else writing to that file which causes the error, shouldn't we wait a bit to let the other thread finish it's job?

[tgodzik]

Seems reasonable, though I'm only wondering if the reindexing doesn't happen to fast. Let's say you really have a something else writing to that file which causes the error, shouldn't we wait a bit to let the other thread finish it's job?

That might be right, but this is happening during resolving a completion item, so we can't wait. We could instead just not retry and the retry might happen next time when resolving an item. What do you think? Remove the retry? It might not be useful at all

[kasiaMarek]

What do you think? Remove the retry? It might not be useful at all

I have no intuition on usefulness of this retry. However, I don't see how it would be harmful.
You could inquire the user if the exception happened once or several times and if it caused any actually disfunction, if not then I also think it's normal for such things to happen. It's your call.