Zero Handle error

[weiweihuanghuang]

I'll send file to you.

[jenskutilek]

There is nothing wrong with the curve. The "zero" handle test actually flags a handle which is in the range of 0 to 2 units from the on-curve point. For very short curves like in your example, this gives a false result. I'll see if I can take the curve length into account for the tolerance range.

[weiweihuanghuang]

Ah, is something that is 2 units from the on-curve point still a problem?

[weiweihuanghuang]

Also if a off-curve point is the first point this happens

[jenskutilek]

Ah, is something that is 2 units from the on-curve point still a problem?

Not really, but for "normal" length curves (I'm thinking half x-height distance), a 2 unit handle would probably mean there's something fishy.

Maybe I have put two things into one outline test that should really be two different test items. Zero handles are a real problem (especially TTF conversion algorithms seem to choke because they cannot compute the angle of the outline at the on-curve point from a zero vector).

I could add a "handle relations are plausible" warning for the other case (one handle long, one very short).

[jenskutilek]

Also if a off-curve point is the first point this happens

Strange, I didn't know it's possible to make an off-curve point the contour start. I think that's not technically possible in an exported font, so perhaps it's a glitch in Glyphs. I'll ask Georg.

[TrueTyper]

Hi Jens,

I know the tool is called Red-Arrow, but I was thinking Jens that a 1 or 2 unit handle could be Yellow (or closer to orange because it is easier to see) to represent a warning rather than error. Just a thought.

Hey, congratulations on the birth of your child!! I hope you are enjoying it immensely, and managing to get some good sleep from time to time. :-)

Tschüss!

Tom

On Sep 3, 2015, at 9:26 AM, Jens Kutilek <[email protected]mailto:[email protected]> wrote:

Ah, is something that is 2 units from the on-curve point still a problem?

Not really, but for "normal" length curves (I'm thinking half x-height distance), a 2 unit handle would probably mean there's something fishy.

Maybe I have put two things into one outline test that should really be two different test items. Zero handles are a real problem (especially TTF conversion algorithms seem to choke because they cannot compute the angle of the outline at the on-curve point from a zero vector).

I could add a "handle relations are plausible" warning for the other case (one handle long, one very short).

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/5#issuecomment-137465546.

[CatharsisFonts]

I know the tool is called Red-Arrow, but I was thinking Jens that a 1 or 2 unit handle could be Yellow (or closer to orange because it is easier to see) to represent a warning rather than error. Just a thought.

In fact, I was wondering whether there could be a few more colors coding for the kind of error. For instance, I'm currently trying to iron out "incorrect smooth connection" errors — it would be good to be able to distinguish them from the more common "extremum" errors that I'm currently ignoring (for things like the tips of slanted serifs etc.).

[jenskutilek]

I’ve made the Zero Handles test more strict in f4a0638. I’m leaving this open, because of the additional ideas in this thread.