fetch_remote_content tool now can get http and gemini content.

With these changes fetch_remote_content command naively parses url schema and either fetches http and gemini content or returns an error.

I'm not sure about the required python changes in here https://github.com/epilys/sic/blob/ed0c162a5700d5e41cf66258356c9d8f2d8b8446/sic/jobs.py#L33

@epilys could you take a look at this branch and verify that nothing is broken?

sic's nameserver, hosted by njal.la apparently has some issues with AFNIC the french TLD registrar (.fr/.re etc) which also manages .pm domains. It's been 5-6 days at the time of writing and still no resolution.

Right now it looks like comments are limited to 500 characters in length. Some restriction is good for the database and making sure users don't write their novel inline. But I ran into the 500 character limit super quickly when trying to reply to a post.

It looks like in the Lobsters rails schema the db field used is a MySQL medium text which would mean there's a limit of 16,777,215 bytes. A 16+ megabyte comment is pretty excessive. I looked through the code and couldn't find any other restrictions on it, though it wouldn't surprise me if I missed something.

If left up to me I'd make the limit 2,000 characters (or really 2,048 because I'm a nerd), since I'd want to encourage more indepth discussion. But figured I'd raise this issue here for discussion. There might be downsides to inlining values that large into sqlite but probably not for a while?

This PR adds a setup.py and a corresponding .gitignore entry to make pip install -e . work. Having that reduces the amount of noise from pylint for me.

It also adds two new commands to manage.py: dump_site and populate_site.

dump_site writes users, stories, tags to JSON and populate_site restores them. This should be a little more resilient to local differences and schema additions than the sql dump has been.

If title field in the submit form is empty, auto fill it with url title upon form submission.

location in codebase

Submitting a story is handled in this view:

https://github.com/epilys/sic/blob/4ca05eea446d01da79288ef90c4270ac06268396/sic/views/init.py#L206-L215

problems

• Field is "required" in HTML, the html attribute must go and instead have the title presence validated in the server response.
• Need to timeout properly when attempting to fetch a title so as not to slow down server.

I’m messing with this from a tablet, so I’ve made a couple changes that make life easier for all that entails:

1. add a requirements.txt to make it easy to spin up a short-lived venv
2. add local_settings.py so I can easily set ALLOWED_HOSTS without putting them in source controlled files… tablet can’t really run a local dev server
3. Ignore .vscode directory because that’s the only sane way to do anything on this tablet, without an ESC key.
4. Ignore dev instance SQLite database because I’ve almost accidentally committed it thanks to stupid vscode’s over-aggressive attempts to add it.

Right now, this repository is being developed on and https://github.com/epilys/tade is getting the new commits by rebase. The site instance should move to the [tade] codebase by creating a new django site that uses tade as a django app.

Enable users to suggest tags for stories they didn't post, and automatically set the tags somehow

User should have a minimum standing in order to be able to suggest tags.

Perhaps there could also be a hidden "suggestion" score that indicates how often a user's tags suggestions are accepted by moderators or story posters and use the score in order to automatically set tags.