[Refactor DPIUtil] rename scaleDown and ScaleUp methods to pixelToPoint and pointToPixel respectively

[ShahzaibIbrahim]

In DPIUtil we use

• scaleUp -> point to pixel
• scaleDown -> pixel to point

My suggestion is to use better naming and change it to what it actually does.

@akoch-yatta @fedejeanne @HeikoKlare if you have any objections/concerns, please mention here.

[HeikoKlare]

I am fully in favor of improving those names to be more comprehensive.
I only see two potentials concerns:

• People might be used to the existing name so they need to get familiar with the new names again. However, it might still be a relief for their brains to not be concerned with doing the on-the-fly matching of a non-comprehensive name to what it's doing all the time by just giving it the information with an according method name right away 🙂
• There might be usages of the existing methods at 3rd party consumers (even though access is discouraged as the class is not exported by the bundle). Those usages are of course invalid but still you have to weigh the benefit of change against potential breakage. I see the benefit in changing names here (also given that the DPIUtil class has been highly refactored/split up anyway).

[fedejeanne]

I see pros and cons to doing it so I would like to have a bit more details about why you propose to do it. Anyone reading the body of the methods could question this proposal on the very simple grounds that the body contains variables like scaleFactor (which speaks about scaling). Maybe you could simply add some usages of this method to support your statement of (quote) what it actually does as preparation for the PR too.

Regarding usages by 3r parties: DPIUtil is internal so technically there is no need to go through the deprecation process. You could mention the renaming in the release notes though.

[ShahzaibIbrahim]

My point is that: the implementation detail (using a scale factor in body) doesn't clearly convey the intent of the method from its name. When someone (me) reads scaleDown(x) or scaleUp(x), the natural question is:

“Scaling relative to what? What coordinate system or unit is this about?”

That’s ambiguous without diving into the implementation.
In contrast, renaming to pixelToPoint(x) and pointToPixel(x) makes it immediately clear. We’re converting between pixel units (physical/screen space) and point units (logical/UI space). The direction of conversion is unambiguous.

and @HeikoKlare about the usages by 3rd party, even if that's the case that some of the methods are used by them. They will break anyway since we have moved most logic into platform specific DPIUtils and almost all of the scaleDown/up methods are now not present in DPIUtil anyway.

[HeikoKlare]

I am still in favor of doing the proposed change. Sorry if it was not clear from my consideration that it's just a collection of things to consider but still I conclude to do the change.
Are there any objections or should we move this to ready?

[fedejeanne]

It's clear now, thank you @ShahzaibIbrahim .

Ready