Inconsistencies between turnstiles and their slashed variants

[anderslundstedt]

As always, thanks for a great font!

I have noticed that the slashed variants of various turnstiles do not look exactly as their corresponding turnstile plus a slash, which I guess would be the desired looks. I mainly use '⊢', '⊣', '⊨', '⫤', but I guess the problem would apply to other turnstiles as well. The turnstiles which have no separate code point for their slashed version look OK when appending u+0338.

[cormullion]

Ah interesting, I will investigate this. One of my skills is being inconsistent... :)

Thanks!

[anderslundstedt]

FWIW, this sort of inconsistency does not seem limited to only turnstiles. For example, the strokes in '≈' gets thinner in '≉', and similarly for '<' and '>' when turned into '≮' and '≯', respectively. On the other hand, many other symbol pairs look consistent to my eyes—for example, '=' and '≠', '≡' and '≢', '≺' and '⊀', and '≻' and '⊁'.

[cormullion]

I made a few changes so that the slashed versions are more consistent (well in this picture even the slashed ones get slashed, which probably doesn't happen in real life):

Sometimes there are many different ways to be consistent - and the fixed width is an added factor to contend with here. It's all a compromise ... (that's my excuse anyway 😃).

[anderslundstedt]

Looks good! Will try them out in the next release.

[anderslundstedt]

Some issues remain:

• 22a2+0338 looks like 22ac and not as a slashed 22a2. (The mirror of 22a2 is 22a3 and 22a3+0338 looks correct.)
• Connected to the previous point, 22ac should decompose as 22a2+0338 (https://www.compart.com/en/unicode/U+22AC) so it should probably be changed as well.
• 2ae4 and 2ae4+0338 look smaller than their respective mirrors 22a8 and 22ad (=22a8+0338).

[cormullion]

I've made a few more adjustments - they're currently on the current master...

[anderslundstedt]

The only inconsistency I now notice is that 22a2+0338 and 22ac do not look like a slashed 22a2 (but more like a slashed 22a6).

[cormullion]

What needs fixing? I'm getting quite confused now... :)

[anderslundstedt]

It might be me misunderstanding something, but here is how I see the problem. Currently 22a2+0338 and 22ac both look like a slashed 22a6. I would like 22a2+0338 and 22ac both to look like a slashed 22a2.

FWIW, in support of my opinions:

• The mirror of 22a2 is 22a3 and 22a3+0338 looks like a slashed 22a3. Thus similarly 22a2+0338 ought to look like a slashed 22a2.
• According to https://www.compart.com/en/unicode/U+22AC, 22ac "decomposes" as 22a2+0338. Thus 22ac ought to look like 22a2+0338, which, by the previous point, ought to look like a slashed 22a2.

[anderslundstedt]

It might be me misunderstanding something, but here is how I see the problem. Currently 22a2+0338 and 22ac both look like a slashed 22a6. I would like 22a2+0338 and 22ac both to look like a slashed 22a2.

Expanding on this: in your image, the length of the horizontal stroke makes 22a2+0338 (=22ac) look more like a slashed 22a6 than a slashed 22a2.

[cormullion]

I think something's not working quite right here, I've been trying to get predictable results for an hour now and failing... Could be a software error, who knows...

I may have to leave this to one side for a while.

[anderslundstedt]

Fair enough!