[ jEdit-users ] Fwd: Errors showing in status bar

[Rowan C. <row...@gm...>]

Sorry, broken mail headers somewhere - I expected "reply" to reply to list,
not author...

---------- Forwarded message ----------
From: Rowan Collins <row...@gm...>
Date: 26 May 2010 18:08
Subject: Re: [ jEdit-users ] Errors showing in status bar
To: Mike <xan...@go...>


2010/5/18 Mike <xan...@go...>

> There should be two types of error:
>
> 1) Expected error: This should ALWAYS be dealt with in a user-friendly
> way... i.e. if the auto-save feature tries to save my untitled document in a
> unix folder where I don't have permissions then a dialog (as you said)
> should say "Can't autosave 'untitled 1' to /usr/bin.  Click here to turn off
> autosave for untitled files, OR you might think about running jEdit as root
> if you need to save a file in that locaiton".  That's just an example error
> message but it's far more constructive, as it:
> a) Tells the user exactly what has happened in clear terms.
> b) It tells them possible options to improve the situation without needing
> to Google or go online!
>


An expected error can only be handled if someone has written the code to
expect and handle it. If they haven't, the error is by definition *
unexpected* - so it will propagate, and become a system error. If a
condition that *should *be being handled by a friendly message is instead
creating a system error, that is a bug, and should be reported. If a system
error is occurring that *cannot* be replaced with a friendly message - well,
that's a bug too, surely?

In other words, fatal errors should always be reported as bugs if you can
reproduce them. If the resolution to the bug is to show the user a friendly
message - or even ignore the error and recover automatically - that is still
a worthwhile bug report.

Now, whether we should expect all "ordinary" users to see and deal with the
fatal errors, I'm less sure. As Shlomy says, automatically filing error
reports is not trivial, and sifting through them to find actual bugs even
less so. I agree with you about making it easier to report bugs in Free
Software ("Freeware" means something different, BTW), though, with caveats
again about sifting effort.

Just my 2-pence worth...
-- 
Rowan Collins
[IMSoP]