Feedback from early show-and-tells

We did a few rounds of early show-and-tell, including:

• this session in a meeting of the Fediverse Developer Network;
• this session at FediForum;
• and a few one-on-one sessions with potential users.

The main points of feedback were:

1. Ah, interesting, is it done yet?

   Comment: I take this as something positive :-)

2. You are doing something complicated (supporting several nodes in a server constellation). Can’t you start by doing something simpler, such as simply running a few curl commands against a server in the cloud?

   Comment: the complex can address the simple (e.g. single server), but it’s not viable to extend the simple to the complex (e.g. multi-server constellations) where all the bugs lurk that today, practically, aren’t findable. We want to cover the entire spectrum.

3. FediTest should provide some value immediately to a new developer after a very quick installation. It can provide more value later after a more complex setup.

   Comment: that sounds smart, let’s see what we can do.

4. Implement TAP. Or something that is like TAP that is easily machine-processable.

   Comment: Sounds like a good idea.

5. For the “test all” scenario, produce a table with application conformance similar to what is often done for web browser features, e.g. see here.

   Comment: That may be a bit more complicated here, because tests tend to involve at least two application nodes, but the idea seems good.

6. I need it to test X.

   Comment: file an issue here https://github.com/fediverse-devnet/feditest-tests-fediverse/issues so we can keep track of it.

Also, nobody objected or proposed something else than my hack-up-the-spec annost hack. Admittedly nobody has been particularly enthusiastic either, but people see the advantages it has for traceability. So we’ll go with that for now.

Have more feedback or questions? Post with a mention to @feditest.