Please stop saying .target is recommended. We would like it to be but it just can't be since: It's slow It needs to download everything on every new workspace It often fails Having the tp locally and use as direct platform works much better since It's faster I just need to point to the target dir on every workspace It doesn't seem to fail(probably because I don't need to download everything every time ;) /max (sent from my phone)

On 02/06/2014 09:19 AM, Max Andersen wrote: This seems to be a misconception. I let you try the following scenario: At the same time * start "mvn clean verify -Dmultile-to-repo.includeSources=true" on jbosstools-target-platforms/jbosstools/multiple * Import TP in workspace and click "Set as Target Platform", or (often better) go to Preferences > PDE > Target-Platform, select the new TP and click "Apply" The first one to finish is declared winner. From my place, IDE takes about 15 minutes to have TP loaded (I tried 3 times), whereas I started mirroring with Maven 50 minutes ago and it's still running. That's true. But since mirroring takes more than 3 times more time, then it's the same time to update TP in 3 different workspace. There is no time saved by mirroring TP with Maven except if you're using it in more that 4 workspace. I wish you would report it when it happens, so we can find out the reason and hopefully fix it. So I confirm and insist that I still (and will probably always) recommend using directly .target in IDE, because it's simpler and more efficient. https://github.com/jbosstools/jbosstools-devdoc/blob/master/building/targ... -- Mickael Istria Eclipse developer at JBoss, by Red Hat <http://www.jboss.org/tools> My blog <http://mickaelistria.wordpress.com> - My Tweets <http://twitter.com/mickaelistria>

Yes, multiple only. The one that is recommended to use for development in IDE as it provides source "naturally", and which is used by Tycho to create the mirror. Why do you need a mirror? Tycho caches so much stuff that you already have a mirror without even configuring one. PDE relies on p2, which relies on ECF, allowing parallel downloads among other optimizations. I guess It also fetches only p2 artifacts and unpack them locally, dividing by 3 to 4 the amount of downloaded bytes. It's just smarter than the mirror goal (that is not part of Tycho, but a plugin of ours). So broken connection are not at all related to how you consume TP. They happen for both "mvn verify" and IDE the same way. This is not a criterion to tell one thing is better than the other. FYI, I also often have broken connection when mirroring TP with Maven. It's not better for that. If you feel comfortable with TP, and want to hack stuff, it's up to you. But beware that several people in the team didn't fully understand it yet and get confused by opposite instructions. The simplest approach is the one documented in https://github.com/jbosstools/jbosstools-devdoc/blob/master/building/targ... , which as for only drawback compared to mirroring locally to make workspace not usable for the time it fetches artifacts. But it's simple, people understand it, and it only cost 15 minutes per workspace every 6 weeks or so. I believe there aren't many arguments remaining for not recommending simply using .target file in IDE :P -- Mickael Istria Eclipse developer at JBoss, by Red Hat <http://www.jboss.org/tools> My blog <http://mickaelistria.wordpress.com> - My Tweets <http://twitter.com/mickaelistria>

On 02/06/2014 07:06 PM, Denis Golovin wrote: Please define "broken" (in a Jira ideally). I've updated documentation to prefer using Preferences > PDE > Target-Platform > Select TP | Reload instead of using "Set as target platform". I have the impression this works just fine (and better). That would be nice if you could try that and provide feedback. -- Mickael Istria Eclipse developer at JBoss, by Red Hat <http://www.jboss.org/tools> My blog <http://mickaelistria.wordpress.com> - My Tweets <http://twitter.com/mickaelistria>