Proposal: Continue the displayException program

[Michael Snoyman]

As I'm guessing you'd expect, I'm +1 on this.

Several years ago, Michael Snoyman proposed [*] that we add a
displayException method to the Exception class, and that we make GHC use it
in the default handler. The first part of that proposal was accepted and
implemented, but the second part was not, for reasons that are not very
clear to me. I would like to re-propose that we change the default
exception handler to use displayException instead of Show. This seems like
a much better default, and anyone who wants to do it the old way can still
change the top-level handler.
[*] https://mail.haskell.org/pipermail/libraries/2014-November/024176.html
_______________________________________________
Libraries mailing list
http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries

[Eric Mertens]

It looks like the original objection is that using displayException makes it harder for the programmer to determine how to handle the exception and that uncaught exceptions were expected to be mostly seen by the developer. This kind of problem certainly comes up when we have to consult the GHC sources to map IOError error message text back to the correct predicate function to use to test for an exception.

Part of the original discussion on this topic included some ideas for ways to make it easier to install an uncaught exception handler that would use displayException instead of the default show. Maybe it would be a good time to entertain some progress on those?

Several years ago, Michael Snoyman proposed [*] that we add a displayException method to the Exception class, and that we make GHC use it in the default handler. The first part of that proposal was accepted and implemented, but the second part was not, for reasons that are not very clear to me. I would like to re-propose that we change the default exception handler to use displayException instead of Show. This seems like a much better default, and anyone who wants to do it the old way can still change the top-level handler.
[*] https://mail.haskell.org/pipermail/libraries/2014-November/024176.html <https://mail.haskell.org/pipermail/libraries/2014-November/024176.html>_______________________________________________
Libraries mailing list
http://mail.haskell.org/cgi-bin/mailman/listinfo/libraries