On Tue, Jun 4, 2013 at 3:06 PM, Chip Childers <chip.child...@sungard.com> wrote: > On Wed, Jun 05, 2013 at 12:22:02AM +0530, Rohit Yadav wrote: >> On Wed, Jun 5, 2013 at 12:14 AM, Chip Childers >> <chip.child...@sungard.com>wrote: >> >> > On Wed, Jun 05, 2013 at 12:09:21AM +0530, Rohit Yadav wrote: >> > > On Wed, Jun 5, 2013 at 12:05 AM, Chip Childers < >> > chip.child...@sungard.com>wrote: >> > > >> > > > The problem I'm seeing is that -SNAPSHOT is a newer version than -0, so >> > > > 4.1.0-0 (the one I published from the release) isn't being downloaded. >> > > > Instead 4.1.0-SNAPSHOT-3 is the one that's pulled. Perhaps I'm just >> > not >> > > > configuring things right? Knowing your schedule, if you don't mind >> > > > taking a look and letting me know if perhaps I've done something wrong, >> > > > that would be helpful. >> > > > >> > > >> > > Yes, that's the expected behavior. To download the release version >> > 4.1.0-0, >> > > one should do: pip install cloudmonkey==4.1.0-0 (or suitable version now >> > > and in future). >> > > >> > > In my view, one should get the latest and greatest (almost stable) >> > > cloudmonkey and choose specific versions using == while installing. >> > > >> > > If you want to enforce that, please go ahead the delete it. We can in >> > that >> > > case ask users to get the latest from the git repo. >> > > >> > > Cheers. >> > >> > Yeah, but the SNAPSHOT version numbers represent incremental work >> > towards the final release. I'm cool with the SNAPs. I'm just thinking >> > that perhaps we need a scheme that allows the actual release version to >> > be the "latest" if it is, in fact, the latest. >> > >> > Anyone else have thoughts on the numbering scheme? >> > >> > Unfortunately Pypi isn't the same as maven, in respecting the SNAPSHOT >> > text in a certain way. >> > >> >> Yes, it tricky the problem will persist if I upload a 4.2-snaphost and >> won't allow the 4.1.0-0 to be downloaded. So, let's do one thing we want to >> make the pypi release channel to sync with the actual ACS release so let's >> remove all snapshots and ask users to get the latest from git repo. For >> snapshots we can have a parallel release channel called >> cloudmonkey-snapshot or something we can make where we can go ape crazy >> with latest releases (maybe everyday or week :), or people can go on create >> their own distros of cloudmonkey with different name and features :D (I >> once tried clouddonkey as the pkg name to release snapshots :P) >> >> It's fine either ways and I trust your judgement Chip so please go ahead >> and enforce whatever behavior we want out of pypi, it's just a release >> channel. >> >> Cheers. > > OK, I've removed -SNAPSHOT from pypi. Let's consider breaking out > cloudmonkey into it's own ASF release artifact... should be easily done > actually. Then we can release on whatever schedule for that project > that is warranted.
I like that idea - and honestly would prefer it be broken into it's own repo to make that easier. I wonder if we don't need to do that with Marvin as well. --David