calt and liga tracking fail

Nick Shinn

The Contextual Alternates and Ligature feature stop working in InDesign when tracking is increased beyond +20.

This is a problem for pseudo-random typefaces in which the effect should work whatever the letterspacing.

Is there another on-by-default OpenType feature I can put this effect in?

It seems strange to publish versions with different built-in letterspacing, but that’s what I intend to do. 

I think that’s a better solution than putting the options in Stylistic Sets, as those are so annoying to locate in GUIs.

Opinions?

And would “Loose” and “Extra Loose” be good names or just “…+50” or whatever the amount is?

Chris Lozos

Lord save us from software that tries too hard to think for us.  It would be just dandy if Adobe would have a preference setting that would allow users  to both turn it off and adjust parameters.

Kent Lew

Nick, there are now the {rclt} Required Contextual Alternates and {rlig} Required Ligatures feature tags.

One would think that software would respect the “required” aspect of these specific tags and leave them on regardless of tracking.

I haven’t actually checked to see if Adobe does this [yet]. If so, I would expect it may require the World-Ready Composer (as I believe these feature tags were added primarily with non-Latin in mind).

You could run some tests for yourself.

Nick Shinn

Thanks Kent! 
That’s a priceless piece of information.
I should have thought to check the chart at Typoptheque.
https://www.typotheque.com/fonts/opentype_feature_support

{rlig} isn’t affected by tracking changes in InDesign, at least as far back as v7.5

However, it still fails when baseline shift is applied to part of a line (although not the whole line), and when point size is changed for less than a line.

Now I have to go back and update all my old [calt} fonts. Grrrrrrr.

Looks like Photoshop and Illustrator don’t support {rlig}, which is unfortunate.

Nick Shinn

After a little bit of testing, I’ve come up with a best practice of putting the pseudo-random code in both the {rlig} and {calt} features, identical (except for renaming the lookups), with {rlig} first.

The good news is that in the major Adobe apps (CS, haven’t checked CC), tracking can be adjusted without breaking the effect. However, in Photoshop and Illustrator, changing Baseline Shift or Point Size of one character in a line (but not the whole line) turns off or disrupts the effect.

Erwin Denissen

I suggest you use the same lookup for both features. It saves space and is easier to maintain.

Nick Shinn

I tried that, but got a ‘fatal’ EOF when compiling.

There are several lookups in the code. 

Erwin Denissen

It is perfectly legal to use a lookup within multiple features, but I don't know what software you use.

Feel free to send me the font file, and I'll run it through FontCreator so you can test it without duplicating the lookup.

Kent Lew

Yes, you should be able to simply reference all of the same lookups from the first feature in the second — basically registering the same set of lookups to each feature tag. Perhaps the error had some other cause that got fixed incidentally when you changed the lookup names.

More importantly, however, I wonder if there would be any compounding issues when both features are simultaneously activated. Perhaps it won’t matter if final changes in the earlier feature don’t result in any targets when the series of subs is repeated. Or maybe the result will just be “extra” randomness. ;-)

Did {rclt} not give nonbreaking results? Or is it not supported widely enough? I would think it would be parallel with {rlig}. I don’t know if the fact that it is not listed in that Typotheque list is indicative of anything (that list seems slightly out of date anyway).

Nick Shinn

Thanks Erwin, I will do that soon.

Kent, I tried {rclt} in InDesign CS and it didn’t work.

As for tow identical features with different tags, that would only be an issue in InDesign, as Illustrator and Photoshop don’t support {rlig}, and InDesign seems to work fine with both {rlig} and {calt} in the font—toggling {calt} on/off makes no difference.

Kent Lew

That’s too bad that {rclt} is not supported as well. And good that your features don’t have any problem with doubling up. (Although, like I said, with the purpose of randomness in this case, it might not have been such a terrible thing. ;-)