Should be a way to skip video encoding entirely if the test suite passes

[kaiyoma]

What would you like?

There should be a way to skip video encoding entirely if the test suite passes. I know that there is a videoUploadOnPasses configuration option, but this only applies to Dashboard, which we don't use. As per #2522 there is a way to use the after:spec event to delete video files for passing suites, which is a bit helpful in terms of freeing up disk space, but is still wasteful in general.

Why is this needed?

Compressing and saving a video file, only to delete it a second later, is wasteful of time, memory, and processing power. There should be a way to skip this process entirely.

[csvan]

You can set compression to 0 to skip the encoding process, which should save some CPU time.

[kaiyoma]

You can set compression to 0 to skip the encoding process, which should save some CPU time.

That sounds like a potentially worthwhile temporary hack. I'll try it out!

Edit: Of course, this can create other problems when tests fail and the resulting videos take up a lot more disk space.

[jennifer-shehane]

I was just revisiting this issue and a couple things to note as workarounds:

• You can set videoCompression: false to skip compression of all videos, this would save time on a test run since compression takes time, as noted in the "Processing" output of a test.
• videoUploadOnPasses, while this option appears that it should only apply to runs with the --record flag to the Cloud, in actuality this flag also skips the video compression for runs that are not recorded to the Cloud. You could pass this flag and skip compression of videos that are from passing specs, then delete them in an 'after:spec' hook.

I recognize this would not be at all obvious from the naming, but I did want to offer what we found while experimenting that could be a solution today.

[csvan]

@jennifer-shehane is there any functional different between setting videoCompression to false or 0?

[emilyrohrbough]

@csvan videoCompression set to false and 0 behave the same. This was corrected in Cypress v12.11.0.