On 2021-09-04 4:28 p.m., Niels Sascha Reedijk wrote:
> I get what you are going for, but the real usefulness here is really
for a developer debugging an application, at which stage you might as
well attach a debugger and instruct it to break whenever an exception is
thrown, so that you get the full stack trace. For general error handling
within a regular application, these backtraces should not be necessary.
(Unless the exception is not handled, after which the application
could/should terminate with a stack trace).
I'm thinking more of end users being mystified by the one layer error
messages. For example, "Package manager failed to install X." is a one
level message. Layers of messages would have that at the top, but under
it, "Dependent package Y failed to install." and under that "File Y.hpkg
failed to download." and under that "URL
https://haiku-os.og/downloads/Y.hpkg not found". With that set of
messages, not quite the same as a stack trace, the user could figure out
that they have a typo in their repository name (og instead of org).
Something that single level messages rarely help with. Using a debugger
to examine the stack would be an alternative for programmers, but
wouldn't be as good for the end user - cluttered with too much other
information and you need to use the debugger.
So, layered error messages: better than a single level error message,
more useful to the end user than a stack trace.
- Alex