This repository captures the steps and scripts them so that a clean migration can be run in a few hours, if and when a migration is ever approved by the PSC.
The migration will encompass the following steps:
- Create blank Github repository/project
- Convert SVN to git and push to Github
- retain branch and tag information
- Convert trac tickets to github format
- retain revision links (r1234) as hash links
- map ticket references (#1234) to github ticket references
- map trac users to github users
- map milestones and versions to github tags
This reads an SVN repository into a Git repository and then runs some specific commands to get rid of un-wanted PostGIS branches and tags. When complete it generates a text file of revision/hash references to that the Trac tickets can be re-written when they are migrated.
Quick and dirty command to pipe the PostgreSQL trac database into a local instance for easy querying.
Really just runs the trac2github.py script, where all the logic is.
The CONFIGURATION section needs extensive customization for any particular repo load.
- Load tags and milestones from repo.
- Convert trac wiki syntax into markdown.
- Add comments and attachments.
- Use the new Github "beta" Import Issue API to support post-dated comments and issues.
Note that the Import Issue API runs asyncronously, so to find out the status of any given import, a secont REST query is required, for example:
curl -X GET -H "Authorization: token xxxxxxxx" \
-H "Accept: application/vnd.github.golden-comet-preview+json" \
https://api.github.com/repos/pramsey/postgis-gh/import/issues/1420269
Frequently a import will be accepted but the content will be rejected later because:
- Issue assignee not in the repository organization (cannot assign issues to people you don't "know", as it were)
- Milestone unknown, label unknown.
Unfortunately the ownership of all issues and comments will be that of the user doing the loading, usually a synthetic user, like the @mapserver-bot.