Raised the following so this doesn't get lost:
http://jira.codehaus.org/browse/MNG-1648
Mark
On 05/11/05, Tomislav Bodor <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I've attached a partial canonical paths patch. I don't know who this
> should really be going to if it's to find its way into the tree, but I
> reckon someone here will know.
>
> This doesn't quite fix the problem you are talking about in this thread,
> but it does fix a similar one that I have encountered. Feel free to grab
> any parts of the patch if you find them useful.
>
> The problem I ran into was related to using a flattened multi module
> project hierarchy, where the pom (master) project is at the same level as
> its modules. The modules are then referenced like this:
>
> <modules>
> <module>../module-1</module>
> <module>../module-2</module>
> </modules>
>
> Without the patch the plugin generates unnecessary links in the .project
> file since it doesn't realise that /some/dir/master/../module-1 is the
> same path as /some/dir/module-1. By using canonical paths for comparison
> (that's all that the patch does), this problem goes away and normal source
> directories without links are created.
>
> Please let me know if I can provide more information.
>
>
> Regards,
> Tomislav
>
>
>
> > On Fri, 4 Nov 2005, Mark Hobson wrote:
> >
> >> Hi Kenney,
> >>
> >> On 04/11/05, Kenney Westerhof <[EMAIL PROTECTED]> wrote:
> >> > I didn't realize the exact difference. I think I added the
> > 'canonical'
> >> > versions, and the 'absolute' versions were implicit. They should
> >> > ofcourse be the same.
> >> >
> >> > Is this correct:
> >> >
> >> >
> >> > /path/a/b/
> >> >
> >> > /path/c symlinked to /path/a/b/
> >> >
> >> > - absolute -> just makes a path absolute, preserving symlinks:
> >> >
> >> > cwd = /path/c; absolute('target/') -> /path/c/target/
> >> >
> >> > - canonical:
> >> >
> >> > cwd = /path/c; canonical('target/') ->
> > /path/a/b/target/
> >> >
> >> > ?
> >>
> >> Yep, that's right. On Windows the canoncial path also converts the
> >> drive letter to uppercase, which is the root cause of the test
> >> failures.
> >>
> >> > I don't know which version would be the correct one. I think
> > maven2
> >> > internally uses Absolute paths to resolve $basedir, so I guess
> > the eclipse
> >> > plugin should use that too.
> >>
> >> Moving to one or the other would fix things, so for the eclipse plugin
> >> we should really use whatever eclipse itself uses. A quick test shows
> >> that eclipse creates canoncial paths - shall we move to that instead?
> >
> > +1, but I think we also need to keep in mind to convert any paths that
> > maven2 itself offers (if they're absolute).
> >
> > -- Kenney
> >
> >>
> >> Mark
> >>
> >> ---------------------------------------------------------------------
> >> To unsubscribe, e-mail: [EMAIL PROTECTED]
> >> For additional commands, e-mail: [EMAIL PROTECTED]
> >>
> >>
> >
> > --
> > Kenney Westerhof
> > http://www.neonics.com
> > GPG public key: http://www.gods.nl/~forge/kenneyw.key
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> > http://mail-archives.apache.org/mod_mbox/maven-dev/200511.mbox/[EMAIL
> > PROTECTED]
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
