A follow-up.

I have finally found some time to properly set the server up anew.
Did so. It works as expected. A test fails for some reason, though. 
I currently don't know why. Luckily, it's a different test, so we
have some progress here!

My advice is to not run tests that depend on the network.
In check(), you can replace

    make test

with

Hello Simon!

First, thank you for working on the package! It means a lot.

Second, bouncepaw.com is indeed down. The server that hosted it got hacked a few days ago. It was wiped because of that, and I have to set up everything again.

I have all the data backed up. I plan to restore the website as soon as possible, but I lack time and energy to do so these days. Thanks to you notifying me of the issue, I'll prioritise it higher now. Expect it this weekend, I hope.

In retrospect, it's an obvious misjudgement on my side to write tests like this. Did I really expect bouncepaw.com to have little downtime? Naïve!

In the meantime, maybe turn off the testing?

Thank you! Applied.

Thanks for the contribution, arne!

Having a User-agent is something we indeed want to have. We have a
ticket [1] for that!

I can't accept the patch in its current form though. The way User
agent is implemented is perfectly fine, I don't like the version
related stuff. You can split this patch in two parts: one user
agent related, one version related, and maybe just ditch the latter
one.

What I don't like about the version:

1) The file storing the version is called .version_string.txt. I do

Took the closest look at the timestamp. It is different from what it was
before, but turns out to be fully standards-compilant! Applying the
patch, thank you!

Wonderful, thank you! I especially like how you bumped x/crypto and
x/net dependencies.

There is an issue with timestamps though.


> +		PubDate:     now.String(),

You turn it into string with String, which ends up with a timestamp like
2024-02-20 10:23:30.466972 +0300 MSK m=+65.648189751, but previously it
was Tue, 20 Feb 2024 07:23:19 +0000. I suppose this format is called
RFC-822. Use the Format function instead.

Meanwhile I'll fix the tests...

Thank you! Ran a quick test, seems to work as expected. Applying. I will provide
the translation myself.

LGTM. Thanks!

Cool feature!

> It works, but the UI is sorely lacking: columns (and the table overall) is too
> thin. `SELECT * FROM Posts` type of queries would probably also benefit
> from displaying them as post feed instead of a table.

Yeah, this is true. In the body block of template, maybe move the <table>
outside of <main>, so it can be as wide as the page is? Or make it horizontally
scrollable? Displaying as posts would be great, but it would be fine to
implement it later, in a different patch. Looking at column names, and deducing
if the result can be shown as a post by that? Sounds doable.

Some logging would also be nice. Also, running an empty query results in Betula
hanging. The database gets locked, maybe?

Thanks! Kudos for writing the test.

I have found a bug though, I think you should fix it and write an additional test
for it. If you save a mailto: link such as betula@example.org, it gets
displayed as mailto://.

I expect a similar problem to happen with schemes like tel: and others that do
not have a slash after the colon.