Gna!

Gna!
URL	https://gna.org/
Status	Closing
Archiving status	In progress...
Archiving type	Unknown
IRC channel	#gnarm (on hackint)

Gna! is a centralized location where software developers can develop, distribute and maintain free (GPL-compatible) software. It is an instance of the Savane code-hosting platform. It hosts for popular free software projects such as Battle for Wesnoth and Freeciv.

It is shutting down due to lack of admin effort; probably by end of April / early May 2017.

Hosted data

As of 2017-02 it claimed to have 1458 hosted projects. (Many are probably abandoned and will not be saved by their project admins before shutdown.)

Here's a breakdown of the kinds of data stored and what various people can do to grab the data:

• Third party describes what random anonymous Internet people (e.g., Archive Team) can do
  • Done shows bits that we have already rescued
  • Help shows bits that someone could usefully do
• Members describes things that only members of the relevant project can do (if better)

Data stored:

• Code hosting using CVS, Subversion, and Arch (done 2017-02-25, not updated since; see subpage)
  • Third parties can grab all code with full history:
    • All subversion repos available via anonymous rsync: rsync://svn.gna.org/svn/ (ref: bottom of every project's svn page e.g. [1]). (In FSFS format, which is supposed to be portable.)
    • Ditto CVS, it looks like: rsync://svn.gna.org/cvs/
    • Arch/tla [2]: rsync://download.gna.org/arch/
  • There's also a ViewVC web front-end to browse code. (No point grabbing this if you've got the above)
• Ticket tracking (not saved, help wanted)
  • Up to 4 trackers per project: 'bugs', 'patch', 'task', 'support'
  • Project members (only) can set up XML export of their own ticket text/metadata ("Export" item on tracker admin menu).
    • Only option for third parties looks like web scraping. Help: can someone look into this? (Someone pointed ArchiveBot at it but it doesn't seem to have grabbed much)
    • Exported XML is published to an unauthenticated URL of the form https://gna.org/export/project/user/number.xml . number might be global; a recent export had number 66. In principle this namespace could be mined by third parties although it's a rather large search space (1458 projects * 9116 users * 66 numbers) and would only catch recent or periodic exports, since they are cleared out quickly.
  • There's no supported interface for grabbing issue attachments (such as patches) even for project admins though.
    • Third parties can scrape attachments by relying on their increasing integer IDs, e.g. file #29845. It looks like you don't have to get the 'bugs' bit correct, so it's possible to scrape all public files by varying the ID. (done/ongoing by JTN, not uploaded anywhere yet)
  • Individual tickets can be private. (Maybe files too?) But the XML export includes private tickets (yes, to an unauthenticated URL).
• File hosting at http://download.gna.org/ (done 2017-02-25, not updated since; see subpage)
  • Third parties can do anonymous rsync from rsync://download.gna.org/download/
• Project websites on home.gna.org (done 2017-02-25, not updated since; see subpage)
  • e.g. http://home.gna.org/freeciv/
  • These are managed via Subversion [3], so grabbing svn by rsync as above should also save website data + history
• Mailing lists using Mailman (not saved, help wanted)
  • Which means public archives are available to third parties in mbox format (albeit with email addresses mangled). e.g. [4] Help: can someone scrape these? Should be easy.
    • Note, the most recent mbox link on inactive lists (e.g., [5]) is broken; replace "2014-09.partial.mbox.gz" with "2014-09.mbox.gz" to fix it
    • It may be worth grabbing the HTML archives too, as they contain some info not available in the mboxes, e.g. "X-From-R13" in HTML comments contains reversibly obfuscated From address
  • Some mailing lists are private. Even project admins can't see the archives at the moment (sr 3421).
• Project metadata: groups, users, news, help topics etc. In a database and probably only available via web scraping. Help: can someone look into this?
• Usage stats at http://stats.gna.org/

Shutdown Notice

• A notice of pending shutdown / request for takeover was first announced in Nov 2016 suggesting a time frame of six months
• A news item about shutdown was posted to the front page 2017-01-31 linking to the above. A reply to that on 4 Feb suggests shutdown will happen "within 3 months, or when the hardware dies".
• This suggests shutdown by around the beginning of May 2017.