Hm, you could have a point! Transitively it might matter, I'm just not sure (now).
On Dec 11, 2007 10:07 PM, Niklas Matthies <[EMAIL PROTECTED]> wrote: > On Tue 2007-12-11 at 09:06h, John Gill wrote on ivy-user: > > That's why we are dependent on a particular version of the > > ivyconf.xml. That way the choice of conflict manager can't change > > when you try and reproduce the build. > > True, it can't change for a particular build, but it could change > between the build and a build of a depending module. Example: > > A ------> C1 -> D > A -> B -> C2 -> D > B -> C3 -> D > > If builds of A uses a different retrieval or conflict resolution > policy for D than builds of B, then B could break (or misbehave in > subtle ways). Having B depend on a particular ivyconf.xml doesn't > translate to A, which could apply a whole different D than B assumes. > > More generally, a published ivy.xml still assumes particular settings > from an ivyconf.xml. These settings must in general be the same > between building some module B and later having its published ivy.xml > being used for building some other module A. So one might want to have > module A have a dependency on the correct Ivy settings for this, but > that (I think) breaks down once the settings change between versions > of module B, several of which might (transitively or intransitively) > be relevant in the resolution/retrieval process for building a version > of module A. > > -- Niklas Matthies > -- Regards, John Gill
