On Thu, Oct 21, 2010 at 2:06 AM, Stephen Connolly <[email protected]> wrote: > On 20 October 2010 16:36, Phillip Hellewell <[email protected]> wrote: >> But if it can be inferred while the plugin is running, why exactly >> does it need to get written back into the pom? Of course, I don't > > You did see the -Dprovider=svn
Yes. > There are N providers not all of them can be inferred... what do we do > if somebody has a svn workspace in a git workspace with .gitignore's > containing .svn and svn:ignore on the root folder containing .git? > That would allow both scm's to function at the same time without being > aware of each other... which should maven choose? Don't infer the SCM. Just infer the URL. I think it's ok if the SCM has to be provided on the command-line (as you illustrated) or inside the pom where the <connection> looks like "scm:svn:infer". > Oh and this git/svn workspace could also be an accurev workspace (no > control files needed to be ignored there) and why not add CVS into the > mix? Yeah, I think it would be great if over time we could add infer ability for more SCMs. >> mind it being written into the pom on a tag, and in the pom that goes >> to the repo, but in the trunk and branches the whole idea is to not >> have it present in there so it is always inferred. > > but you are going to have to specify at least the provider... so at > least more than nothing will have to go in there... plus when that > -SNAPSHOT gets deployed wouldn't it be nice to know from which branch > it was deployed? oh look the pom has the scm details of the branch it > came from... cool! that was joe's branch which explains why I'm fecked Yeah, I'd be fine having <connection>scm:svn:infer</connection> in there. I'm also fine with having it get replaced in the pom in the tag and in the released poms that gets deployed, but in the trunk/branches it could stay as scm:svn:infer forever. The only downside I can see is what you just mentioned about snapshots. I agree it could be helpful to know like what branch the snapshot came from, but I still think having infer functionality has some great benefits that for some people would outweigh this one drawback. Phillip --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
