You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The purpose of a rescue_from block is of course to handle exceptions in a custom way. However, what should happen if the rescue_from block itself raises an exception? Currently, Grape does not handle such exceptions, likely resulting in an Internal Server Error returned by the web server.
In some cases, it would be beneficial if an exception raised inside a rescue_from block was handled by another rescue_from block. However, it’s unclear whether that’s a good idea in general, as it could result in an infinite rescue_from block loop.
I have a use case where Grape::Exceptions::ValidationErrors should be remapped to a different exception that is a subclass of Grape::Exceptions::Base. What I currently do is the following:
However, it no longer works with newer Grape versions (probably since #2377).
For this use case, it would be best if it was possible to raise an exception in the rescue_from block that is then handled by the default rescue handler for Grape exceptions. If that’s an option, I can try to come up with sensible semantics that allow that while avoiding infinite loops.
Otherwise, I’d welcome suggestions for handling this use case cleanly and such that it works with current Grape versions. The best I’ve come up with is to call #error!, however it can be a bit fiddly to convert the exception to the correct arguments for #error!. Note that this requires a recent fix (#2471).
The text was updated successfully, but these errors were encountered:
If you throw :error the code will handle the case but not a raise. It's quite easy to handle it.
Putting throw :error, invalid_value_error in the rescue block in the above example doesn’t work with or without #2483. It causes the middleware to return an Invalid Response error. The line error_response(response) is not executed in this case because #error? returns false for non-Hashes.
In case of an exception in the rescue_from block, with your code, it passes the exception to #default_rescue_handler which always returns HTTP error code 500. If it called #error_response for subclasses of Grape::Exceptions::Base, it would work.
The purpose of a
rescue_from
block is of course to handle exceptions in a custom way. However, what should happen if therescue_from
block itself raises an exception? Currently, Grape does not handle such exceptions, likely resulting in an Internal Server Error returned by the web server.In some cases, it would be beneficial if an exception raised inside a
rescue_from
block was handled by anotherrescue_from
block. However, it’s unclear whether that’s a good idea in general, as it could result in an infiniterescue_from
block loop.I have a use case where
Grape::Exceptions::ValidationErrors
should be remapped to a different exception that is a subclass ofGrape::Exceptions::Base
. What I currently do is the following:However, it no longer works with newer Grape versions (probably since #2377).
For this use case, it would be best if it was possible to raise an exception in the
rescue_from
block that is then handled by the default rescue handler for Grape exceptions. If that’s an option, I can try to come up with sensible semantics that allow that while avoiding infinite loops.Otherwise, I’d welcome suggestions for handling this use case cleanly and such that it works with current Grape versions. The best I’ve come up with is to call
#error!
, however it can be a bit fiddly to convert the exception to the correct arguments for#error!
. Note that this requires a recent fix (#2471).The text was updated successfully, but these errors were encountered: