On Sun, Jan 23, 2011 at 4:49 PM, Mike Edwards <mike.edwards.inglen...@gmail.com> wrote: > On 22/01/2011 14:25, ant elder wrote: >> >> The problem with doing that is that now the OASIS repo is also checked >> for every artifact when building online which could add significantly >> to the total build time. >> >> Looking at the current repository definitions i think most of those >> are old historical ones which we don't even need anymore so i'm going >> to have a go a cleaning them up. For the ones we do still need i think >> we should change to define them in the modules that actually require >> them instead of the top level pom.xml as that should avoid them being >> queried for irrelevant artifacts. >> >> Besides all that why does that itest need to be using the oasis >> version of the API's? The Tuscany version should work fine shouldn't >> it (i tried it, it seems to)? >> >> ...ant >> > Ant, > > I think the boot is on the other foot. > > The SCA API jar is really one that belongs to OASIS - why does Tuscany have > its own version of that jar, now that it is available from a Maven repo? > > I understand that Tuscany had to have its own API module historically, but > why do that now? > > > Yours, Mike. > >
I guess the main reasons in favour of using the Tuscany version is that it gives us control more control over changes. It would also be better if the OASIS version could be added to the Maven central repository as maven people really dislike things not being in the central repo. Whatever happens I think that should be done as a conscious decision by the project and if the decision is to use OASIS then change everywhere consistently, having the Tuscany build mix the use of both the OASIS and its own Tuscany version of the API classes doesn't make a lot of sense, and in this itest case both these jars are now in the classpath at the same time duplicating the classes. ...ant
