Re: [tei-council] I've just test-migrated my CodeSharing project from SF to GH

Hello, I'm customizing the script that Martin experimented with to get as much information as possible transferred over, such as adding labels for feature requests, bugs and "milestone" colours. I'd like your feedback on how we should keep track of the ticket creator and ticket assignee. As we discussed during our last council call, the imported issues are all created by the user importing them. We can set up a TEI-C user that will own all of the imported issues. The original assignee can be assigned to the imported issue only if they have a GitHub user (I could locate the GitHub user of 46 out of 88 contributing SF users). Otherwise the issue can be assigned to no one, everyone (any), or a default user. My question is: where should we keep track of the original owner and assignee? They could be stored - in the text body of the issue - as labels, e.g. sf_assignee:some_user and sf_owner:some_user Using labels would allow us to filter and search these imported issues effectively. But we'd end but with a large amount of labels (at most 88 for ownership plus 42 for unmatched assignees). And those labels should not be re-used after the importing process. Thoughts? Raff On Thu, Jul 30, 2015 at 6:19 PM, Raffaele Viglianti < raffaeleviglianti@gmail.com> wrote:

Hi Martin, You beat me to it: I was working on exactly the same stuff with the same tools :)

Raff On Jul 30, 2015 6:18 PM, "Martin Holmes" wrote:

...

Hi all, especially Raff, our migration manager.

I just migrated my project from SF to GitHub in the following steps:

=== 1. PREPARATION ===

Do a freeze (not a problem for my tiny project).

Make sure you have complete backups and checkouts.

Prepare a JSON file called users_sf2gh.json which consists of the equivalent user names from SF and GitHub, like this:

{ "martindholmes" : "martindholmes", "stadlerpeter" : "peterstadler" }

(Obviously this needs a bit of research.)

=== 2. GETTING TICKETS ETC. OUT OF SF ==========================

Do a full export of the SF project from here:

https://sourceforge.net/p/[proj]/admin/export

This is not instant; you'll get an email when it's complete and you can download it as a zip. That produced these files for me:

admin.json discussion.json tickets.json wiki.json

It doesn't include the code, of course.

=== MOVING CODE TO GITHUB =====================================

Go here:

https://import.github.com/new/

and follow the instructions to import your code. This was quick for my tiny project, but would take a long time for the TEI codebase. The commit history is preserved correctly, as far as I can see. I provided the https URL for the SF repo:

https://svn.code.sf.net/p/[proj]

While that's going on, or after it's finished, you're offered the option to provide the list of SF/GitHub user equivalences you've already specified in the JSON file, through a web form (you'll use the JSON file itself later, when importing tickets).

=== MOVING TICKETS TO GITHUB =================================

I used this PERL script for this:

https://github.com/cmungall/gosf2github

Beware: there is a typo (occurring in two places) which omits a quotation mark in the example; I've added a patch and sent a pull request to fix it.

There are several stages to this:

1. Set up an OAUTH token for the Git repo here:

https://github.com/settings/tokens

2. Get a list of the project collaborators in JSON format from github like this:

curl -H "Authorization: token YOURTOKEN" https://api.github.com/repos/[entity]/[project]/collaborators > [project]-collab.json

3. Run the PERL script, which pulls down each ticket from SF and pushes it up to GitHub -- this is my command line:

perl gosf2github.pl -a martindholmes -u users_sf2gh.json -c codesharing-collab.json -r martindholmes/CodeSharing -s codesharing/tickets -t [MYTOKEN] tickets.json

You can run it with the -k flag first to do a dry-run and see the curl statements that would be addressed to github without sending them.

This successfully imported the two tickets I had created as a test, one open and one closed, with embedded XML code successfully preserved.

Hope this helps. It's nice to be on vacation. There's time to actually do stuff. :-)

Cheers, Martin

-- tei-council mailing list tei-council@lists.tei-c.org http://lists.lists.tei-c.org/mailman/listinfo/tei-council

PLEASE NOTE: postings to this list are publicly archived