rel-html: change default file extension to tar.xz

The kernel.org server is now not dumping out tar.bz2 files
so embrace their practice for preference for relying on tar.xz
tarballs. You can always override the extension with the
release_extension attribute, see the project file for hostapd
projects/hostapd.cfg as an example.

Signed-off-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
1 file changed
tree: 5a9ee2e53a1ec2373a635830998f043730cba28a
  1. html/
  2. projects/
  3. agpl-3.0.txt
  4. CONTRIBUTING
  5. LICENSE
  6. README.md
  7. rel-html.cfg
  8. rel-html.py
README.md

rel-html

rel-html is designed to parse naked index html pages with tarballs on a software project and automatically produce a nice shiny HTML5 release page for you. It takes as input a configuration file, rel-html.cfg, in which you can specify attributes for the release. Exact stable releases are inferred based on some hints, but we still require at least the base releases that are supported to be annoated.

Example release page

TODO

  • Figure out how to automatically determine releases from git.

    • If we have many stable releases how should we annotate this via git ?

    It seems that the way to go is to require a config file for the project with the oldest stable release supported and then annotate eols. We do this right now, so rel-html would just need to be modified to infer newer releases.

  • The Linux kernel now (as of 2013-03-10) has json file for releases:

    https://www.kernel.org/releases.json

We need to do a few things then:

- Get other projects to use json for releases
- Add json intepreter support to rel-html

If projects don't use json releases files as the Linux kernel does then the current usage of HTMLParser would allow us to parse / infer releases for us.

  • See if we can copy the EOL release into an eol/ directory and moving forward instead of parsing the tags use the directory name to automatically determine other release attributes. This is only relevant for the Linux kernel right now.