ant-ivy-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joshua Tharp <joshua-th...@alumni.calpoly.edu>
Subject Re: [Request] Help with ivy:publish - bad revision found in ivy file.
Date Mon, 25 May 2009 19:42:23 GMT
That's a good theory. Here's the docs on the update property of publish:

*update*


true to update ivy file metadata (revision, branch, publication date and
status) before publishing, false otherwise. This is usually not necessary
when using deliver before publish. No. Defaults to false
So I perhaps if you use the deliver ant task before publish, then you
wouldn't have this problem. Otherwise, use the update = true attribute on
the publish task.

Glad you found it.

Josh

On Mon, May 25, 2009 at 12:28 PM, Stephen Nesbitt <
stephen_nesbitt@alumni.cmc.edu> wrote:

> Joshua:
>
> Thanks. That wasn't it, but your notes led me to see something I did miss.
>
> I *think* what is happening is the following:
> 1) I run a publish-local on a completely clean instance. A new revision id
> is
> generated, ivy does a resolve and places an ivy.xml with my revision id in
> my
> out.dist directory everything works.
>
> 2) I run publish-local a second time on an existing instance. A new
> revision
> id is generated, Ivy does a resolve, but *nothing* has changed - a new
> ivy.ml
> file is *not* generated but the old one retained. We now have a mismatch -
> I
> generated a new revision id, but the ivy.xml has the old one. Boom -
> publish
> fails.
>
> My theory here rests on the assumption that Ivy does not generate a new
> local
> ivy.xml file unless it has to - it either doesn't exist or there have been
> no
> changes that would warrant a new ivy.xml
>
> -steve
>
>
> > (Disclaimer: I'm not an expert either, but since I'm also working on a
> > holiday, I'll take a shot...)
> >
> > What does your local repository pattern look like? Is the revision in it
> > someplace? If so, what is the revision number on the directory name? Is
> it
> > 1.4 or your longer ${ivy.revision}-<timestamp>-<build type> string?
> >
> > My guess is that according to Ivy you are trying to republish revision
> 1.4
> > a second time (that the other parts of your name are not being used by
> Ivy)
> > and it has found the first copy in your local repository. So, Ivy is
> > recommending that you either force a delivery (to overwrite the first
> 1.4)
> > or update it (to only change certain fields in the published ivy file).
> >
> > My other suggestion is to check to see if you really want ivy.revision
> > (which is the last revision number found in the repository) or
> > ivy.new.revision (which is automagically incremented to a new build
> > number).
> >
> > Good luck!
> >
> > Josh
> >
> > On Mon, May 25, 2009 at 11:17 AM, Stephen Nesbitt <
> >
> > stephen_nesbitt@alumni.cmc.edu> wrote:
> > > All:
> > >
> > > I could really use some help figuring out how to successfully perform
> an
> > > ivy:publish. Here's the scenario:
> > > 1) Everything including local cache cleaned. We are starting completely
> > > clean.
> > >
> > > 2) As part of my build initialization, I create a build identifier. It
> is
> > > composed of the ${ivy.revision}-<timestamp>-<build type>. For example:
> > > 1.4-20090525100000-dev. I see an ivy.xml file with the correct
> identifier
> > > in my
> > > out/dist directory.
> > >
> > > 3) I perform a publish to my local repository. Everything works just
> > > fine. I
> > > now have an ivy.xml in my local repo with the expected build identifier
> > > in the
> > > revision attribute.
> > >
> > > 4) I perform another publish *without* cleaning anything. The
> ivy:publish
> > > task
> > > errors out with an "impossible to publish artifacts...bad revision
> found
> > > in ivy file... Use force deliver or update" error message referencingf
> > > the ivy.xml
> > > in my out/dist directory
> > >
> > > Huh? This makes no sense. Even if I try with an update it fails? I am
> > > obviously missing something.
> > >
> > > Can someone tell me what is going on and how to fix it? Sure I could
> > > update and
> > > force deliver, but that doesn't feel right and I don't understand it.
> > >
> > > What is the correct way to set the revision on a publication?
> > >
> > > I will be more than happy to write up the fix, if someone would tell me
> > > what it
> > > is :-)
> > >
> > > Thanks
> > >
> > > -steve
> > > Stephen Nesbitt
> > > Absaroka Tech
> > > Build & Configuration Management Consulting
> > > Seattle, WA USA
>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message