Downloads for website via a repository

Issue #152 closed
René Dudfield created an issue

Instead of using the 'Downloads' part of bitbucket for downloads, it makes sense to use a repository instead.

This way syncing with the website can be done more easily, and it also makes it easier for the buildbots. The 'Downloads' section does not have an API currently, which makes dealing with them hard with scripts.

Also, some minor points are that this allows people to submit downloads via a pull requests, and it also provides a history and MD5 sum for the downloads.


  • create new 'downloads' repo under the 'pygame' team.
  • move all old downloads into this new 'downloads' repo.
  • update release documentation, and hacking documentation to reflect this.
  • website script which generates downloads webpage based on update information from hg 'downloads' repo.

Comments (11)

  1. René Dudfield reporter

    Perhaps there should be a separate repo for build bot builds than releases?

    Also, the date stamps and the files should be the same if possible, so that mirroring code that checks will work still.

  2. René Dudfield reporter

    Generating the html should also be done. We need other meta data for each file. So that should be stored and edited somehow. A downloads.json file will do for now.

    What goes on the download page exactly needs to be considered. See the existing download pages, and other websites download pages.

    It needs to fit in a small column (for mobiles etc).

    The new website system works with jQuery and json. So a html page which takes a downloads.json, and uses jQuery to populate a bit of html would work.

  3. René Dudfield reporter

    There are install tips scattered in many places around the internet, the wiki, and the issue tracker. For example in the wiki, 'Compilation' parts.

    Things like homebrew on Mac, raspberrypi, and other platforms should be in a "Getting started" type page.

  4. Paul Craven

    For the last point, how is the script to be created? Will it be run as a cron job on a server, or javascript to get the info live?

  5. René Dudfield reporter

    @pcraven : There's a script which generates html from the JavaScript, and json. Yeah, a cron job of sorts. Except it watches for changes on the file system and does the business.

    @techtonik : wheels sound usable definitely. I believe there's still problems with the wheel format for linux distros, but it works at least for windows and Mac OS X. On OS X, this may involve moving the Framework dependencies inside of the pygame package. Rather than having them install into separate locations. I'm thinking this is the right approach for the .dmg OS X installers too. Here's the wheel docs... Issue See #222 for tracking wheels.

  6. Log in to comment