This is just adding tests, not running them, right?

Yeah, this doesn't run them through TravisCI. @keneanung at some point will hook them up I imagine, I'll worry about making them and running them locally.

This looks like a very nice tool! There are some niggles right now:

• I needed to manually tell squish where the Mudlet executable is and there is no was to include that in the project files. We should document the how or at least refer new users to https://doc.froglogic.com/squish/6.2/rg-autsettings.html#rgas-pathmapped
• Using strict screenshot comparison made the tests fail on my machine because there were minimal (like 0.057% and 0.0006%) differences. We'll probably want to use a threshold of around 0.1-0.5% for differences.
• In one test run, I had an additional newline in the tst_2_make_triggers test, which made the test fail. I wonder why and if we can be a little more flexible here...

Yeah, you have to map the executable.

I ran into the same issue with screenshot comparison mode - I made a little script to prune identical images and ended up with this set of images:

I think a solution to this would be if we could tell squish to look for an image anywhere within an image - instead of just a static mask that blocks out other parts for comparison. That would solve our problem. I haven't found out how yet though nor written to their mailing list about it.

Oh yeah, though one obvious solution is to have a profile that doesn't connect to a MUD - just putting in example.com:23. Then we should have a static output - so we'd solve the problem in this case but not necessarily for other cases.

Does this allow to check for certain colours in the text?

Vadim wrote:

Oh yeah, though one obvious solution is to have a profile that doesn't connect to a MUD - just putting in example.com:23...

I just use localhost:23 ... 😀

@keneanung via screenshot comparison

I need to work on this branch while also having access to the improvements done in #955 @keneanung anything you'd like me to do on this branch?

I'll be swapping out Avalon.de with a blank profile that the BDD creates as a setup hook so we can avoid the connection timing issues - just used that profile as a starting point because it was there already.

Nothing else right now. We may need to decide about BDD or scripted tests and some other guidelines when we have a little more experience with the tool, but I don't think it's something we can decide right now.

Okay, tick please

Ahh sure, my bad. I will add it in first!

@keneanung added. Haven't migrated other tests to BDD yet to make use of the Squish profile - that'll come later.

I think we should go BDD because it's a heck of a lot easier to understand tests written in English than in code for test newbies, and I'd like to get any interested people in contributring - so the bar's gotta be pretty damn low.

I'll leave porting the other tests to BDD later, want to get started on testing #955.