How do you inspect a UFO file?

Josh Nychuk

In Finder I see only the .ufo file, and I'm wondering how to inspect or open a UFO in order to access the contents as listed on Unified Font Object website.

To clarify my goal at the moment, I'm trying to check the file size of kerning.plist and back-check against what is considered an appropriate file size. Bonus points if you throw in what is considered normal file size for Adobe Latin 3 Character Set  smile:

Software:
macOS
RoboFont
MetricsMachine

Jack Jennings

You can right-click the file and select "Show Package Contents" to get at the files inside the document bundle, or you can drop the UFO file into your favorite code editor (Atom, Sublime, etc.) to edit the file contents.

Josh Nychuk

@Jack Jennings thank you!

Kent Lew

Not sure how checking the file size of kerning.plist gets you anywhere useful. The size of the source kerning data dictionary does not necessarily correlate very directly with compiled size and performance. A lot can depend upon the nature of that kerning data.

A seemingly modest kerning data set can explode considerably during compilation.

See also this thread: http://typedrawers.com/discussion/960/number-of-kerning-pairs-vs-performance/#Comment_12465

Josh Nychuk

@Kent Lew thanks for mentioning this. It was that thread actually that made me raise the question, as I'm unsure how to be aware of final font file-size, during the kerning process.

This is a first-time, self-taught project so learning as I go and trying to be aware of potential errors that would involve going back over a lot of work.

Based on your comment how do you go about checking/balancing source kerning data size with the compiled size, so that it's not too large? also not sure what 'too large' would be?

Kent Lew

You keep kerning and compiling until you get the dreaded “kern overflow” warning and then you go back one step and stop. ;-)

Seriously, though, if one is working on a large glyph repertoire that requires a lot of kerning, then one either becomes educated on the technical details of GPOS kerning and adopts strategies/priorities accordingly, or one engages an experienced font technician to help resolve issues with large kerning sets that are not compiling.

In terms of the former, unfortunately, it’s not easy to boil it all down to a forum post.

But you’re in NYC, right? — if I make it down to Typographics in June and you can find me, buy me a beer (or two) and I can try to share some of what I’ve learned. ;-)

Josh Nychuk

Thanks Kent, I didn't actually realize compiling/checking was part of the kerning process...very helpful.

I am in NYC yes, and will be at Typographics. I'd love to take you up on your offer, so let me know if you make it!

Kent Lew

That was tongue-in-cheek. If one is experienced, it is not necessary to keep compiling to check. As you gain experience, you learn how to construct judicious classes and what to kern (and what not to bother with).

I would say that beginner mistakes are usually the result of either not getting the character fitting well in hand before kerning, having too-liberal kerning classes, or attempting to kern everything with everything else.

It’s not always wise to try to generalize about such things, but I will say, just for a frame of reference, that in my experience a regular-weight serif text font with Extended Latin language coverage, with small caps and two sets of figures, competently fitted and reasonably kerned will average somewhere on the order of 3500–4000 kern pairs.

Others may have a different experience.

Don’t take such numbers too literally. That’s mostly just a check on order of magnitude. Less doesn’t necessarily mean a font isn’t kerned well. A bit more than that doesn’t necessarily mean there’s anything wrong either.

Josh Nychuk

Knowing what and what not to kern is basically where I find myself. I've chosen to design a low contrast sans-serif text face, so the spacing is loose...if that's the correct term.

I think I'll begin with kerning all the obvious issues in the uppercase, move to numerals and punctuation and finally uppercase/lowercase pairs and see what i get. 

After kerning 10 pairs or so I've gone back to fixing a few curves, and then side-bearings (which really helped character fitting on some of the rounder shapes)...although I realize this isn't proper workflow.

The advice on averages and balancing metrics with optics is very helpful, thanks!

Mark Simonson

The best practice is to get the spacing (adjusting the left and right sidebearings on the glyphs) to look as good as possible without kerning. Only after you are satisfied with that should you start kerning. Kerning is meant to deal with exceptions—spacing of glyphs that can't be addressed by adjusting sidebearings alone. Usually, these are glyphs with left or right profiles that are uneven, like T, A, L, J, r, period, comma, etc., and often glyphs with round profiles.

Something that novice type designers often get tripped up by: Kerning is not the same as spacing. Kerning is a system for handing exceptions to normal spacing, which is dictated by the sidebearings.

Josh Nychuk

Thanks Mark for mentioning the importance. I've been working with Emil Ruder's list of problematic words from Typographie: Manual of Design but if you have any other sources for testing I'd love to know.

Thomas Phinney

When necessary, FontLab VI automatically optimizes kerning at compile time to avoid overflow, by judicious use of rearrangement and subtable breaks.