feat: update java-errors.data
#4113
Conversation
|
📊 Quantitative test results for language: |
|
Ping! |
|
Ping @fzipi |
|
I don't see much difference, as we're including the "exception name". |
|
So if you happen to have a site that shows java documentation, you surely need to disable this. By adding extra prefixes, wouldn't that prevent some basic use cases? E.g. previously having the Can you add this concept to the prompt you used for generating this list, to see if we get additional strings to match? |
|
If you check line 2 and 50, you'll see the errors are similar to the ones I've added. I don't see any reason for FPs, unless a site shows java documentation itself, in which case normal java code will also be suspect to FPs. |
|
Using the same tool you used for generate this, I get this: I still think adding the |
|
I think it's a good idea. Adding prefixes is simple with regex-assembly. |
We're still waiting for the results on performance comparison between pmFromFile and regex. |
|
For this case the difference will likely be negligible since the list is rather short. |
|
Actually, there are a lot of exceptions: https://docs.oracle.com/en/java/javase/21/docs/api/search.html?q=exception and errors: https://docs.oracle.com/en/java/javase/21/docs/api/search.html?q=error, and I've only scratched the surface. |
|
But you don't have to list every exception. Since you're using a regular expression, you can shorten, e.g., |
|
I'd proceed. I don't see that analysis being done any time soon. |
…nto update-java-errors
|
@Xhoenix Let's switch to regex then? |
|
Waiting for review. |
|
Looks good, tests are failing. |
Co-authored-by: Max Leske <250711+theseion@users.noreply.github.com>
|
Thanks @Xhoenix! |
Closes #4073.