Find consistent way to refer to vertical mirroring

[Enivex]

We seem to only have two current conventions:

.rev - Horizontal mirroring
.inv - Rotate by 180 degrees

These can also be combined to yield

.rev.inv- Vertical mirroring

There are several drawbacks with the current situation:

• The word "invert" can also imply vertical mirroring. In fact, this is the sense in which it is used in unicode naming (though inconsistently), with "turned" used for rotation by 180 degres.
• If there is only a vertically mirrored variant, then using .rev.inv isn't exactly the most natural modifier. This is the case for obtuse angles, where ⦦ (currently angle.oblique) and ⦧ (not currently named) are the only variants.
• Even when there are other variants, it would still be beneficial to have a single modifier for vertical mirroring for semantic purposes.

I see several possibilities

• Introduce .turned, .flip or .rot (though this has also been proposed for 90 degrees) for rotation by 180 degrees, and use .inv for vertical mirroring
• Keep .inv for rotation by 180 degrees, and use .flip for vertical mirroring
• Something like .rev.h and .rev.v (though I can imagine this leading to conflicts if we don't have Allow repeated modifiers to simplify naming #94 Allow the use of order-dependence to disambiguate symbol variants #95 )

[MDLC01]

I agree that the .rev.inv situation is unfortunate.

I also agree that .rev and .inv are not necessarily the clearest names, but I feel like they work anyway for two reasons: they are short, and they still convey the general idea. When reading Typst source code, it's probably not that bad if the modifiers don't make it perfectly clear exactly the way a symbol is rotated/mirrored. When writing Typst code this is more of an issue as you need to remember which modifier gives you the symbol you want, but you only need to figure this once per symbol (then I feel like most people can remember the names of the symbols they use frequently).

Regarding your specific propositions:

• .turned is too long imo. Additionally, .flip would have to be .flipped for consistency (Use "slanted" equal instead of "slant" equal #43 (comment)).
• To be clear, that would just mean changing .rev.inv to .flip? If so, I like that. Although there is again the issue of "flip" vs. "flipped".
• I don't like that because .h and .v are widely used with a meaning that applies to the symbol as a whole.

Regarding the past participle issue ("flip" vs. "flipped"), maybe we can accept "flip" as an abbreviation of "flipped", the same way "rot" is an abbreviation of "rotated".

[Enivex]

This is part of the reason why I dislike adding -ed unless it's required for disambiguation. It's making every such modifier two characters longer for questionable benefit.

If we call it an abbreviation then that's fine with me.

[Enivex]

For symbols that are horizontally symmetric, should .inv or .flip be the default way to refer to the upside down variant?

This currently affects three symbols ⸸ ¡ ℧, but more in the future.

To be clear, that would just mean changing .rev.inv

We haven't yet introduced any such symbols, so no actual change is needed. Just a policy, because there are such symbols in unicode.

[MDLC01]

For symbols that are horizontally symmetric, should .inv or .flip be the default way to refer to the upside down variant?

To me, "flip" evokes rotation. So I'd say .inv is better for horizontal symmetry.

[Enivex]

For symbols that are horizontally symmetric, should .inv or .flip be the default way to refer to the upside down variant?

To me, "flip" evokes rotation. So I'd say .inv is better for horizontal symmetry.

But if flip evokes rotation, should we not switch the roles of inv and flip?

[T0mstone]

Worth noting that the Unicode names for our .inv symbols have differing adjectives:

• amp.inv, comma.inv, dagger.inv, semi.inv, iota.inv: turned
• excl.inv, quest.inv, interrobang.inv, Omega.inv: inverted

[Enivex]

Yes that's why I mentioned inconsistently. Turned is (seemingly) consistently rotation by 180 degrees, but inverted can mean either.

[knuesel]

To me a simple "flip" evokes more a reflection than a rotation, probably due to its use in computer graphics and matrix computations (see OpenCV cv::flip, Unity flip methods, Matlab fliplr and flipud, Numpy flip, etc., though of course a 180° rotation can be expressed as a flip around two axes). Also "flipped upside down" is a common expression, so I'd be in favor of using flip for vertical mirroring.