One-shot mode broken in 1.12

[AdamNiederer]

Repro:

$ for i in {1..200}; do redshift -O 6000; done

Expected: Screen filter is set to 6000K
Actual: Screen is totally red

In redshift 1.11, one-shot mode would set the screen filter's intensity to the provided argument regardless of how many times it was called. Now, calling it multiple times without redshift -x in between appears to increase the intensity per-call.

This change wasn't in the release notes, so I'm assuming it's a bug. This happens to both color temperature and brightness (when called with -O ... -b ...)

[vicr123]

Seems like it's not resetting the old redshift setting before setting it again; if you use a value below 6500K and then set a value above 6500K, the screen brightness will start to decrease.

[mattjohnruss]

Git bisect suggests that the issue was introduced in commit dcafe50 from pull request #511 related to preserving gamma ramps by default (this bug could affect other methods as well since this pull covers others).

[mattjohnruss]

Ah, just found this closed issue #513 which seems to be pretty much a duplicate? The "bug" seems to be the intended behaviour and using the "-P" with one-shot mode as @jonls suggests here (#513 (comment)) makes the issue disappear for me.

[jonls]

Yes, using -P should work for this.

[gavtroy]

I might question whether the new behavior is a sensible default in the first place, but the larger issue is that the api was broken without warning and (compared to adding a flag for the new behavior) without apparent benefit.

Users are going to experience this as a bug, regardless of whether it was intentional. Although there is a workaround (-P), it would obviously be preferable to fix it upstream, rather than confusing users and wasting their time.

This is not a good experience. Considering some larger distros have yet to pick up the update (and some users will have held it back, thinking the bug will disappear in the next release), I think it is worth reconsidering how this issue is treated.