Do you guys use Unit Testing ?

[ArthurAttout 1]

This is a genuine question, it has never been in my habits to rant on anything / anyone, and I've never had the intention to make anyone feel bad about their contribution or their work in general.

 

There is something that has been bugging me for a few months now.

Emby is releasing updates and patches at an insanely high frequency.

Standalone components (like Dlna) are updated almost three times a week (at least that's what the dasboard tells me).

 

With such a tremendously high amount of changes, it is obvious that regressions are to be expected.

 

That's what drove me out of the beta-testing channel.

First I thought myself "That's ok, that's the point of the beta-channel, it's supposed to be used as some kind of large-scale testing to reach out for feedback quickly", so I went out, and switched to the stable release channel.

 

As for my experience, there is the same amount of regression in beta-channel than in stable.
And we're not talking about deeply buried functionality that broke : since the update of the Cinema mode on the 23rd of April, the seekbar stopped working.

I couldn't jump to a specific point of a video by clicking on the seek bar.

 

The seekbar stopped working. On the HTML5 interface.

 

I'm not trying to make anyone feel bad about this, but I just cannot understand how this is even possible.

Such regression are just unacceptable, I do not understand how this got passed any sane unit-testing suite, or QA revision.

 

So my question is : do Emby community even bothers using any sort of regression detection mechanism (there are tons of them, plus CI/CD fanciness, etc.) before releasing an update ?

 

Again, I do not want to sound rude, but I think at this scale (probably more than 100K users), regressions like basic seek handling shouldn't ever occur.
 

NB : I've never looked at the Emby Premiere versions, but I seriously hope it's going through more testing and QA before release ...

Edited April 24, 2019 by ArthurAttout