On Jun 16, 2011, at 10:23 AM, Peter Cock wrote:
On Thu, Jun 16, 2011 at 3:26 PM, Greg Von Kuster <greg@bx.psu.edu> wrote:
Hi Peter,
On Jun 16, 2011, at 10:10 AM, Peter Cock wrote:
Are you expecting tool authors to work primarily at the hg level?
You didn't answer that one ;)
Not necessarily, but hg is the basis for uploading and downloading tools. I'm not sure if it will be possible to completely eliminate the requirement of a tool developer using hg, but we're making every attempt to do so.
Good.
Why are you so averse to using hg?
Because git suits me better? ;)
Seriously, I have no strong aversion to hg - what puts me off a little is the need to have one repo per tool or tool-suite, compared to my current setup where all by tools are in a branch from the main Galaxy repo. There would be a significant time and effort cost in switching, made worse by having multiple tools on the Tool Shed.
I'm actually thinking of this (requiring hg knowledge) as a more general issue, namely a potential impediment to new Tool Shed contributors.
Peter
I'm of the opposite mind; I'm not sure there is an absolute need to have one's tools be a branch or fork of the main tool shed, though I agree there is also utility in allowing that (having a customized 'main' repo, or optimizing tools already present). Having completely separate repos for tools seems cleaner, focusing development on those tools alone (not any of the others present in a branch) and pushes maintenance of the tool back to the tool developer themselves (e.g. there is no need for the galaxy devs to 'pull' in changes at any point into a main repo). This might also allow the galaxy devs to designate a set of 'blessed' or supported tools in various repositories. Re: git: as Peter knows I'm also primarily a git/github user. It is feasible at some future point to allow git/github repos. For instance, one could possibly integrate github usage via this: http://hg-git.github.com/ Of course, I think it's much more important that any additional vcs integration wait until the new tool shed interface stabilizes somewhat, but (at least from the github perspective) seems like it shouldn't be terribly hard to do. chris