It is true that pages are a "foreign entity" in a spreadsheet, and that most of a page style's properties only apply to the output (print, PDF...) of the document.
In fact, the only exception to that (that I can see) is the Text Direction: it's the only property that also affects what is displayed in-app, on canvas. And it looks like it's the only cell style property that can use the superordinate value. (Which adds to the fact that linking cell styles to page styles is an odd thing.)

So I don't think this particular issue warrants an overhaul of how styles are handled in Calc.
Maybe the first (and only?) step in resolving this is to dissociate the property Text Direction from Page Styles, so cell styles and page styles are two unrelated things. (A related discussion would be envisaging the creation of Sheet Styles, but I'm not actually sure this adds anything to what we already have with Cell Styles.)

Would you agree to reframe this issue as "remove Text Direction superordinate linking between Cell and Page Styles"?  Or am I missing something?
(And if you agree, then the solution for bug 138862 would be to have a "Default (LTR)" cell style applied to LTR sheets.)

How about:
* cell's Text Direction property has the option "Same as sheet direction" instead of the current superordinate setting;
* Text Direction setting removed from the page style dialog (although this would be problematic for backward-compatibility)

(In reply to Eyal Rozenberg from comment #0)
> This can be verified when examining the effect of changes to Page
> style on properties of cells which are inherited from the superordinate
> object, like text direction; see bug 138862.
Can you think of other cell properties inherited from the page style?

(In reply to Stéphane Guillou (stragu) from comment #1)
> So I don't think this particular issue warrants an overhaul of how styles
> are handled in Calc.

But can't this argument be made in the opposite direction? If there's very little in the way of superordinate-subordinate relations, then isn't it easy to make sheets the superordinate object of cells?

> Maybe the first (and only?) step in resolving this is to dissociate the
> property Text Direction from Page Styles, so cell styles and page styles are
> two unrelated things. (A related discussion would be envisaging the creation
> of Sheet Styles, but I'm not actually sure this adds anything to what we
> already have with Cell Styles.)
> 
> Would you agree to reframe this issue as "remove Text Direction
> superordinate linking between Cell and Page Styles"?  Or am I missing
> something?

I'm not at all sure. And that's because formatting properties of sheets are a thing:

* Default properties of cells within the sheet, e.g. dimensions, and perhaps even a default cell style (named or spelled out).
* The horizontal progression direction of cells
* The order of coverage of cells by pages (vertical major horizontal minor or the other way around)
* What should page margins and borders be on the rim of a sheet and internally?
* Whether or not to print some kind of fallback border between cells which have no border
* Sheet-level header and footer

etc.

Now, you could argue that some of these are not formatting, but actual content. And that's at least partially right. but then - what about the dimensions of a page, for a page style? And why are a lot of these present as part of the Page Style dialog, on the Sheet tab (which doesn't exist in Writer for example)?

And you could also argue that a lot of these only have to do with printing; but that's again also true for Page Styles.