2016.09.07 16:14 "[Tiff] remotesensing.org URLs (http and ftp)", by

2016.09.13 11:32 "Re: [Tiff] Fwd: Re: remotesensing.org URLs (http and ftp)", by

On 2016-09-08 17:01, Edward Lam wrote:

On 08/09/2016 12:26 PM, Bob Friesenhahn wrote:

Longer term we should perhaps decide if we migrate away from cvs.maptools.org + bugzilla to something else (github, osgeo trac with git, etc...), but I guess that we'll have the incentive only when things will break out...

At the moment Github is most popular. It is difficult to predict which offering will succeed and will be the most responsive if some change needs to be made.

Github was my first idea as soon as I heard about this news. It has a lot of infrastructure these days that makes it useful, including the ability to offer a web site directly from committed changes in a special branch of the repo. Other things I like about it are that we can set it up to automatically run builds on multiple platforms whenever a pull request (ie. patch submission) is submitted.

I was originally going to write just this, but since you already wrote it let me second this. A DVCS is the way forward.

That said, Github isn't the only player. There are other alternatives including:

(both broadly equivalent to Github), and then also

I'd have no problems with GitHub (I already use it extensively, and I also use all of the others above to varying degrees). All would offer public git services. But the Github/Bitbucket/Gitlab would also be handy for providing a website since you can host it directly from git. And if ownership of libtiff.org is regained, it can be simply pointed there.

On github, it would be useful to have an organisation for libtiff for developers and contributors to join. If it's possible for the github user by that name to relinquish it, that would make things easy; has anyone contacted them? If not, shall I do that? It looks dead and unused for 2.5 years, unless they also have private repos. Their email is in the single commit in the single repository for the account (assuming it's valid).

If the existing git mirror is of acceptable quality, we could just shift to using it. If there are any niggles relating to username->email identity mappings or branch/tag history conversion, we could redo the conversion from scratch for the migration to any of the git services. I'd be happy to assist with that having done a few previous CVS/SVN/Arch to git conversions for other projects.