Avoid unnecessary catching of Exception instead of StandardError (conver...

[stopdropandrew]

...ting Exceptions into StandardErrors)

Very similar to c989160, but I limited the scope to places where Exception was rescued and then re-raised as a StandardError.

[rafaelfranca]

What is the motivation?

[stopdropandrew]

For me, when sending a SIGTERM or SIGINT while a template is rendering, the process is rescuing the signal and not exiting cleanly.

Rescuing Exception to handle some exit logic (ActiveRecord transaction rollbacks) and the re-raising the original error is fine. But in these cases, an Exception could be rescued and then the code is raising a StandardError, essentially lowering the importance of the error.

Here's some additional info from a similar pull request - #6759

[rafaelfranca]

I'm fine with the changes on the template rendering but I don't think we should change the parameters parser code. If a user submit some data that cause an Exception subclass error it will not handled and the application will die, causing a deny of service.

[stopdropandrew]

That works for me, I removed the param parsing part, the template rendering is where I was actually noticing the issue.

If I start seeing it come up elsewhere I can be explicit about re-raising specific Exception subclasses.