Launching Unstable Antenna Test Instance

I have set up an instance where I from now will be running the latest Antenna code from dev/feature branches, and I would be absolutely delighted if you would try it out. Do you want to try insertion attacks or blocklist evasions? Just see if something is broken, or if new features are available?

This is the place to do it!

Feel free to coordinate with me if there's anything you want me to set up before some experiment.

Of course this instance is not by any means stable. You can expect intermittent CGI Error messages, or things just not producing the expected results. Tell me when that happens, please, and I'll see what I can do about it.

What's New Right Now?

Direct Publishing

That's right. You will no longer have to wait for the scheduled job to run every five minutes to see your thing. As soon as you submit your feed it'll be fetched and published. That means the request takes a little longer as there's a bit to do. On the other hand you will get direct feedback if something fails with your feed!

I tested by submitting Antenna's own atom feed a moment ago, and the whole thing was ingested and published in a couple of seconds.

Safer Block Rule Matching

URLs are now properly cleaned up before attempting to match rules in the blocklist. That means for example that gemini://warmedal.se/~bjorn/posts/../atom.xml will not slip through if gemini://warmedal.se/~bjorn/atom.xml is blocked. That particular case was already accounted for, but there were no doubt other ways to fiddle with the URL to get around it. They should be fixed. Should.

Proper Post Removal

Previously if you had 5 posts in your feed submitted to Antenna and wanted to change or remove 1 you could do so and re-submit the feed. However, and this has as far as I know never actually been attempted, if you decided to remove all 5 posts the page wouldn't actually be regenerated until someone else added posts. Weird, huh?

That's fixed now. If you were to remove your feed file and re-submit it still won't work. I'm going to have a look at that later but my current refactoring efforts, while making the code clearer to read, actually made this a little harder to implement.

Other

Most other changes are invisible as they have to do with codebase organisation and configuration. I've cleared the most important and complicated milestone towards my current goal. Next up is a bunch of cleaning. I don't consider the new code to be production worthy just yet.

-- CC0 ew0k, 2022-05-04