ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steve Loughran" <>
Subject Re: [VOTE] core extensions
Date Sun, 22 Apr 2001 02:40:44 GMT
There seem to be different world views of error handling. There is the
'exception' paradigm, of ML, C++ and successors. There is the 'backtrack'
paradigm of Prolog and similar 'truth seeking' languages. There is the
implicit 'test for failure as you go' paradigm of C. And finally, there is
'On Error Goto'.

Which of these is appropriate for a build/deployment tool, and how can it be
done in a way which makes sense to people who are inevitably versed in one
of the more common error handling paradigms -test for failure and

From: "Ken Wood" <>
To: <>
Sent: Friday, April 20, 2001 11:05 AM
Subject: Re: [VOTE] core extensions

> I believe this was already discussed
> before. If the on fail target fails,
> it would have a "on fail" that is as
> simple as possible, such as a one liner
> that echos "fatal error"...
> I believe the original example posed had a cascading
> a series of on fail targets, each with less
> work (less likely to fail) and each expressed
> increasing severity of the nature of the failure(s).

ok, here's another problem

<target name="main" depends="init" onfail="handler" >

<target name="handler" depends="init" onfail="fail2">

<target name="fail2" onfail="fail2">

<target name="init">

Should the handler for the primary target (main, then handler) direct
failure handling for the dependent task? If not, what happens when a
dependent task fails?


ps: more feature creep. Can we have a mail task that does Mime attachments?

View raw message