b). So I am moving this over to scm-dev ;-).

Ok, here are some of the details.

The problem is actually with the maven-scm-provider-svnexe project. More
specifically, the changelog consumer (SvnChangeLogConsumer).

This class has a couple of regular expressions to parse the header of
the svn log output. This regular expression isn't as robust as it could

The SvnChangeLogConsumer expects the author to be in a very specific
format. Here is a snippet of the author regular expression.


This fits two basic formats for author:

(firstname lastname)



But what happens if the first or last name contains spaces? The entire
header does not match and the changelog returns nothing.

Further, Subversion does not guarantee the format of the author. Hooks
can be added to subversion to modify the author again resulting in an
empty changelog.

The change log consumer should simply consume whatever is in the author
section of the output to keep these kind of mismatches from happening.
This allows for a header match to be found and the scm:changelog command
to produce reasonable output.

So I recommend that we change the parsing of the author to pull in

Todd Thiessen

> -----Original Message-----
> From: Mark Struberg [mailto:strub...@yahoo.de] 
> Sent: Tuesday, January 13, 2009 5:49 PM
> To: Maven Users List
> Subject: RE: SCM plugin project
> Hi Todd!
> Is it a) an enhancement to the maven-scm-plugin itself?
> Or b) does it dig deeper into maven-scm and we have to change 
> the scm-API too?
> For both options you'd probably file a Jira [1] which 
> describes the motivation and the changes you like to do 
> first. Then you may attach a patch or provide your changes 
> via a online repo.
> I personally would prefer git ;)
> For option b) we should definitely switch over to the scm-dev 
> list again.
> LieGrue,
> strub
> [1] http://jira.codehaus.org/browse/SCM

Reply via email to