Return-Path: Delivered-To: apmail-jakarta-ant-dev-archive@jakarta.apache.org Received: (qmail 55659 invoked by uid 500); 7 Aug 2001 13:27:25 -0000 Mailing-List: contact ant-dev-help@jakarta.apache.org; run by ezmlm Precedence: bulk Reply-To: ant-dev@jakarta.apache.org list-help: list-unsubscribe: list-post: Delivered-To: mailing list ant-dev@jakarta.apache.org Delivered-To: moderator for ant-dev@jakarta.apache.org Received: (qmail 51030 invoked from network); 7 Aug 2001 10:31:28 -0000 From: =?iso-8859-1?Q?Magn=FAs_=DE=F3r_Torfason?= To: Subject: Handling build failures Date: Tue, 7 Aug 2001 10:31:21 -0000 Message-ID: <003e01c11f2c$1aa72bf0$6d010a0a@fjolnismenn.local> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook CWS, Build 9.0.2416 (9.0.2910.0) X-MIMEOLE: Produced By Microsoft MimeOLE V5.00.3018.1300 Importance: Normal X-Spam-Rating: h31.sny.collab.net 1.6.2 0/1000/N I have been reading the discussions on the list regarding a failure task/target and saw that such a thing has been rejected. I believe this/target task would be very convenient for a lot of people but understand the arguments that the implications would be to great. I have however been wondering if two additional attributes to the ant task could alleviate this shortage. The attributes would be: failonerror: Defaults to true, to preserve current behaviour, and determines if the current project will fail if the called project fails failproperty: If specified, defines the name of a property that will be defined if the called project fails, but not if it succeeds. This would enable a large build to be put together from small builds, handling failure of subprojects (which may not be critical) appropriately. Also, this would enable special handling, such as mailing the log for the failed build to a responsible person, in an automatic build environment. Forcing people to resort to shell/batch files to handle failed builds, or writing their own build listener, seems to me to be a rather inconvenient solution for handling failed builds, in the long run. This addition would enable build failures to be handled completely in ant instead. Cheers, Magnus Torfason