Intheloop

- Replaced favicon
- New icon for bug tracker
- Typo fix

Freshly introduced, already found a bug. When retrieving the summary of a blog post in the RSS feed, I am getting HTML. I already stripped the tags. But I didn't decode HTML entities. With chinese characters, pretty much everything is an escaped HTML entity and that was recognized as English. So I added HTML entity decoding, too. Language detection works better now!

Wrote a tiny migration script with Supabase + Google Cloud Client in Node

This allows only showing English blog posts by default, otherwise other language will be mixed in. Using Google Cloud Translation API. Quick, reliable and cheap (for me at least), will unlikely exceed the free quota, if so, it'll only be a few cents.

Heavily using JSONB in the Supabase Postgres database. Unfortunately, using a function for sorting does not work in the Supabase JS Client.

I.e. my JSONB column is named info and it has a field createdAt

{"created": "2020-05-05T10:00:15.000Z", "other": "prop"}


With plain SQL, you could do ORDER BY (info->>'createdAt')::TIMESTAMPTZ. That doesn't work via JS client so I introduced a separate field.

Just had to add a book and a few courses

There's no podcasts, newsletters, books or courses. Not many RSS feeds to parse blog posts yet. Not too many communities (not necessarily a bad thing). It's also a pretty fresh technology.

Finished scraping popular tweets yesterday. Added a Firebase function to return a list of tweets for a topic (requerying the Twitter API to get latest information) based on the tweets I marked as popular earlier. Allows me to also include caching when it's needed in the future and avoid any API rate limits.

Took a while because I couldn't receive any mails with my namecheap email and Netlify DNS. Found out over an hour later that I was still using Vercel DNS and forgot to switch.

I was expecting this to be quicker but it took a lot of time to get into the Twitter API. Lots of confusion, my main account not being allowed to register for a dev account and so on. I'm glad I finally got this one to work and any adjustments are fairly easy from now on.

Running scheduled via Firebase and writing the data to Supabase.

This took way longer than expected. There's V1 and V2 (alpha). There's multiple tiers and each tier may be able to use different query parameters or even content within query parameters. The official Postman collection for V2 uses the old OAuth 1.0, while the API does support OAuth 2.0 and some endpoint work with OAuth 1.0, but only when you're using specific parameters, otherwise they only work with a bearer token. This frustrated me a lot more than it should have. Anyways, finally figured it out.

Still need to implement the backend function to actually signup

Next up: Automatic retrieval of blog posts.