Hi Craig, Sorry for the delay. I will not have the time right now to create a simple test case that reproduces this behaviour. What I can do though, is that if you make sure that there are more useful information in DEBUG level traces for these tasks in version 1.7 (like it seems you did), I can analyse these traces when version 1.7 is out (whenever this happens, as we are likely to upgrade).
Mac On 23 fév, 03:01, "Craig Sutherland" <[email protected]> wrote: > Hi Mac, > > I tried changing it to the publishers section but that didn't make any > difference. Obviously this is one of those bugs designed to test the limit > of our abilities :-( > > Craig > > > > > > > > -----Original Message----- > From: [email protected] [mailto:[email protected]] On > > Behalf Of Mac > Sent: Tuesday, 22 February 2011 8:44 p.m. > To: ccnet-user > Subject: [ccnet-user] Re: Conditional and FTP tasks handling ofrelative > directories > > Hi Craig, > > Our configuration is quite big right now, and makes intensive use of > preprocessing. I will try to write a simple test case. > In the meantime, the only difference I can see from your simple > configuration is that my conditional publishing tasks are in the publishers > section (not the tasks one). Maybe that would be a key point for this > problem. > > Thanks for your help. > > Mac > > On 21 fév, 20:50, "Craig Sutherland" <[email protected]> > wrote: > > Hi Mac, > > > Thanks for the additional feedback. I've tried to track down what is > > causing this problem but without any success :-( > > > I have modified the BuildPublisher task to both provide additional > > logging and to explicitly check for the presence of the source > > directory (and to report when not found.) From the looks of things it > > is correctly generating therelativedirectory name, so I'm not sure where > else to look. > > > Here is the configuration that I am using: > > <cruisecontrol xmlns="http://thoughtworks.org/ccnet/1/6"; > > xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance";> > > <project name="RelativeFolders"> > > <workingDirectory>C:\Temp\Working</workingDirectory> > > <artifactDirectory>C:\Temp\Artifact</artifactDirectory> > > <tasks> > > <conditional> > > <conditions> > > <folderExistsCondition> > > <folder>C:\NewFolder</folder> > > </folderExistsCondition> > > </conditions> > > <tasks> > > <buildpublisher> > > <alwaysPublish>true</alwaysPublish> > > <sourceDir>SomeWhere</sourceDir> > > <publishDir>ElseWhere</publishDir> > > </buildpublisher> > > </tasks> > > <elseTasks> > > <buildpublisher> > > <alwaysPublish>true</alwaysPublish> > > <sourceDir>SomeWhere</sourceDir> > > <publishDir>ElseWhere</publishDir> > > </buildpublisher> > > </elseTasks> > > </conditional> > > </tasks> > > </project> > > </cruisecontrol> > > > And I get the following output in the log: > > [RelativeFolders:DEBUG] Checking conditions [RelativeFolders:DEBUG] > > Checking for folder 'C:\NewFolder' > > [RelativeFolders:INFO] Conditions passed - running tasks > > [RelativeFolders:DEBUG] Starting task 'BuildPublisher [0]' > > [RelativeFolders:DEBUG] Publish directory is 'C:\Temp\Artifact\ElseWhere' > > [RelativeFolders:DEBUG] Source directory is 'C:\Temp\Working\SomeWhere' > > [RelativeFolders:WARN] Source directory 'C:\Temp\Working\SomeWhere' > > does not exist - cancelling task [RelativeFolders:INFO] Tasks > > completed: 0 successful, 1 failed [RelativeFolders:INFO] Integration > > complete: Failure - 22/02/2011 8:26:20 a.m. > > > Are you able to provide an example configuration where this problem > occurs? > > > Craig > > > -----Original Message----- > > From: [email protected] [mailto:[email protected]] > > On > > > Behalf Of Mac > > Sent: Tuesday, 22 February 2011 5:02 a.m. > > To: ccnet-user > > Subject: [ccnet-user] Re: Conditional and FTP tasks handling of > >relativedirectories > > > Hi Craig, > > > I made some further tests and it appears that: > > * buildpublisher/sourceDir understandsrelativepathsin general. > > * buildpublisher/sourceDir does *not* understandrelativepathswhen > > used inside a conditional task. This is the behaviour that puzzled me > before. > > > This behaviour is hard to detect, because the buildpublisher does not > > complain (and/or output any trace) when the source directory does not > exist. > > > Seen from (very, very) far away, I would say something unexpected > > happens when cloning the integration result (line 408 in > ConditionalTask.cs,http://ccnet.svn.sourceforge.net/viewvc/ccnet/trunk/proje > ct/core/task... > > tionalTask.cs?revision=7881&view=markup). > > > Mac > > > On 21 fév, 16:22, Mac <[email protected]> wrote: > > > Hi Craig, > > > > Thank you for taking care of this matter. > > > > I was pretty sure I tested arelativepath and that it did not work. > > > But the code seems to be taking care ofrelativepathsalright, so I > > > checked again (with TRACE level logs) and it seems I was wrong. > > >Pathsin the folderExistsCondition *are*relativeSorry for this > > > false alarm. > > > > Mathieu > > > > On 18 fév, 01:35, "Craig Sutherland" <[email protected]> > > > wrote: > > > > > Hi Mac, > > > > > I have gone and updated the Ftp task to userelativepaths- > > > > however it looks like the path in folderExistsCondition is already > > > > arelative > > path? > > > > > Craig > > > > > -----Original Message----- > > > > From: [email protected] > > > > [mailto:[email protected]] On > > > > > Behalf Of Mac > > > > Sent: Wednesday, 16 February 2011 3:16 a.m. > > > > To: ccnet-user > > > > Subject: [ccnet-user] Conditional and FTP tasks handling of > > > >relativedirectories > > > > > Hi all, > > > > > I have just been caught by this feature: almost every task in > > > > CCNet understandsrelativedirectories (from the current working > directory). > > > > For instance the sourceDir property of the buildPublisher task, or > > > > the attachments property of the email task. > > > > > But the following properties do not (I checked the source code to > > > > be > > > > sure) : > > > > * folderExistsCondition in the conditional task. > > > > * localFolderName in the FTP task. > > > > > I got mislead by that, and I suppose other people will/have been. > > > > So I suggest (either) : > > > > * make these properties understandrelativefolders (preferred IMHO). > > > > * or make it clear in the documentation thatpathsmust be > > > > complete for these properties. > > > > > What do you think? > > > > > Mac
