Contrast ratio and rounding

[patrickhlauke]

Following discussion here w3c/wcag#200 (comment) there is/will be an issue with the way in which CCA currently rounds results to single decimal point and then uses the rounded value for its pass/fail determination.

4.5:1 and 3:1 need to be treated as absolute thresholds, and the non-rounded result needs to be assessed against this contrast. so for instance 4.49124578:1 needs to fail as it is below 4.5:1 in absolute terms - it can't be rounded first to 4.5:1 and THEN compared to the threshold figure.

this may require some extra tweaking of the displayed text for these cases. Maybe for values that are between 4.45 - 4.4999999 (or whatever the precision internally is) the output should read "The contrast ratio is: just below 4.5:1" (and the same for 2.95 - 2.9999 for the 3:1 ratio)

xref ThePacielloGroup/CCA-OSX#23

[DavidMacDonald]

Halloween bug
https://twitter.com/stevefaulkner/status/791944961952407552
Mwha ha ha ha

[patrickhlauke]

👻 👻 👻

[ghost]

updated b8445be

[patrickhlauke]

@jun325 tested out the new 64 bit exe. the rounding seems to work perfectly (using #777777 / #FFFFFF as a test combination fails and shows "just below 4.5:1") (but see my comment about default.ini values having regressed?)

[ghost]

Sorry. It's my fault. I have just uploaded correct ini.

[MakotoUeki]

Have you fixed this issue with the latest version 2.4.0.10? It seems that it is still 4.5:1, not 4.48:1 when using #777777 / #FFFFFF with 2.4.0.10. I'm not sure if I uninstalled the previous version appropriately though.

One more thing. Please specify the version number (ex. "2.4.0.10", not "2.4") on the "About" panel (menu [Help] - [About ...]) so that users could confirm the version they are using and they could distinguish the bug-fixed version from the previous versions.

[patrickhlauke]

@jun325 did this regress? confirming that in the latest version, #777777/#FFFFFF still returns a pass / 4.5:1, rather than a fail "just below 4.5:1"

[patrickhlauke]

@jun325 ah, it actually seems you never pushed the fixed version to the stable branch here. are the changes to the rounding good to be added here? once this is done, i'll create a new release package

[patrickhlauke]

The latest release https://github.com/ThePacielloGroup/CCA-Win/releases/tag/2.4.0.23 includes the correct rounding, as well as showing the full version number in the "About" dialog.

[patrickhlauke]

[stevefaulkner]

"this may require some extra tweaking of the displayed text for these cases. Maybe for values that are between 4.45 - 4.4999999 (or whatever the precision internally is) the output should read "The contrast ratio is: just below 4.5:1" (and the same for 2.95 - 2.9999 for the 3:1 ratio)"

although the reporting of the non rounded result 4.45 - 4.4999999 is shown as a fail, the displayed value is still rounded without indication "The contrast ratio is: 3.0:1" so needs to either report the actual result to 2 decimal places or add the clarification text "just below" /cc @jun325

[jake-abma]

Hi @patrickhlauke, any idea when a new Mac version will be launched? At the moment the two versions (Mac / Windows) have different results: https://developer.paciellogroup.com/resources/contrastanalyser/

Thx!

[patrickhlauke]

@stevefaulkner you sure you're on the latest version? it IS showing the "just below" bit in the result for me.

[patrickhlauke]

@jake-abma unfortunately no indication just yet when the Mac one will be updated, sorry. Will discuss with @stevefaulkner

[stevefaulkner]

@stevefaulkner you sure you're on the latest version? it IS showing the "just below" bit in the result for me.

probably not then, will check and close bug if so.

@jake-abma unfortunately no indication just yet when the Mac one will be updated, sorry. Will discuss with @stevefaulkner

Am organising @ferllings to have time to work on it