Bug 10366 - EM DASH (U+2014) should be shorter than HORIZONTAL BAR (U+2015)
Summary: EM DASH (U+2014) should be shorter than HORIZONTAL BAR (U+2015)
Status: REOPENED
Alias: None
Product: DejaVu
Classification: Unclassified
Component: Sans (show other bugs)
Version: unspecified
Hardware: All All
: medium normal
Assignee: Deja Vu bugs
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-03-21 10:57 UTC by Christopher Stone
Modified: 2008-06-29 13:48 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
Comparison of U+2014 (em dash) and U+2015 (horizontal bar) in DejaVu Sans (544 bytes, text/html; charset=utf-8)
2008-06-29 11:31 UTC, Etan Wexler
Details

Description Christopher Stone 2007-03-21 10:57:16 UTC
EM DASH is used in pairs to offset parenthetical text, there is usually a gap in between the two elements.  As of version 2.11, the EM DASH element is just as long as the HORIZONTAL BAR element.  I cannot tell these two apart.  The unicode character charts show that EM DASH (#x2014) is slightly shorter than HORIZONTAL BAR (#x2015) and slightly longer than EN DASH (#x2013).

NOTE also EN DASH (#x2013) appears to be centered, where I see it as being aligned to the left.
Comment 1 Denis Jacquerye 2007-03-21 11:42:52 UTC
(In reply to comment #0)
> EM DASH is used in pairs to offset parenthetical text, there is usually a gap
> in between the two elements.  As of version 2.11, the EM DASH element is just
> as long as the HORIZONTAL BAR element.  I cannot tell these two apart.  The
> unicode character charts show that EM DASH (#x2014) is slightly shorter than
> HORIZONTAL BAR (#x2015) and slightly longer than EN DASH (#x2013).
Horizontal bar doesn't have to be wider than em dash. It can be, but doesn't have to be. Horizontal bar, know as quotation dash, is used for quotations but it is often as wide as em dash. They are two different characters but can be the same glyph.

> NOTE also EN DASH (#x2013) appears to be centered, where I see it as being
> aligned to the left.
En dash currently doesn't have any right nor left bearing, so it is centered.
I don't see why it should be aligned to the left. Could you clarify?
Comment 2 Christopher Stone 2007-03-21 11:52:37 UTC
(In reply to comment #1)
> Horizontal bar doesn't have to be wider than em dash. It can be, but doesn't
> have to be. Horizontal bar, know as quotation dash, is used for quotations but
> it is often as wide as em dash. They are two different characters but can be
> the same glyph.

From what I understand, Horizontal Bar is used to _introduce_ quoted text, whereas em dash is used to _offset_ parenthetical text.

So for example, Horizontal bar would be used like this:

-"To be, or not to be."

and em dash would be used like this:

"To be, or not to be." --William Shakespeare

Note, how em dash is usually used in pairs, and there is supposed to be a small gap between the two dashes.

> 
> > NOTE also EN DASH (#x2013) appears to be centered, where I see it as being
> > aligned to the left.
> En dash currently doesn't have any right nor left bearing, so it is centered.
> I don't see why it should be aligned to the left. Could you clarify?
> 

What I meant was that in sans there is no gap present when you put two em dash characters side by side, there should be a small gap between them.  If you put two horizontal bars side by side there should be no gap.
Comment 3 Denis Jacquerye 2007-03-21 12:16:57 UTC
(In reply to comment #2)
> From what I understand, Horizontal Bar is used to _introduce_ quoted text,
> whereas em dash is used to _offset_ parenthetical text.
Yes.

> "To be, or not to be." --William Shakespeare
> 
> Note, how em dash is usually used in pairs, and there is supposed to be a small
> gap between the two dashes.
You're right about the gap in between, but two em dash that looks wrong.
People use(d) two hyphens to emulate one em dash. 
A double em dash would be super long.
Comment 4 Christopher Stone 2007-03-21 13:10:10 UTC
(In reply to comment #3)
> (In reply to comment #2)
> > "To be, or not to be." --William Shakespeare
> > 
> > Note, how em dash is usually used in pairs, and there is supposed to be a small
> > gap between the two dashes.
> You're right about the gap in between, but two em dash that looks wrong.
> People use(d) two hyphens to emulate one em dash. 
> A double em dash would be super long.
> 

The Unicode Standard, Version 5.0 states "may be used in pairs" for em dash.  Also, EN dash is supposed to be 1/2 the size of EM dash (as far as I understand it).  So as far as I can see, en dash is slightly larger than hyphen, and em dash is twice as big as em dash (but still small enough to have a small gap between two em dash characters). And horizontal line is the largest and has no gap between characters.

I am no expert by any means, I could be completely wrong on this, but this is how the dashes look in the pdf http://www.unicode.org/charts/PDF/U2000.pdf
Comment 5 Denis Jacquerye 2007-03-21 13:37:52 UTC
* r1702 /trunk/dejavu-fonts/ (9 files): * added bearings to en dash U+2013, em dash U+2014 and figure dash U+2012 by making dashes shorter, preserving character width

These three dashes will now have spaces in between when paired. Horizontal bar didn't change.

Thank you for the bug report.
Comment 6 Christopher Stone 2007-12-07 15:29:23 UTC
Hi, I just tested this using version 2.17 and it appears that while this has been fixed with serif, it is still broken with regards to sans-serif.
Comment 7 Etan Wexler 2008-06-29 11:31:42 UTC
Created attachment 17452 [details]
Comparison of U+2014 (em dash) and U+2015 (horizontal bar) in DejaVu Sans
Comment 8 Etan Wexler 2008-06-29 12:10:59 UTC
In reply to comment #0 and comment #4, I quote a disclaimer that appears on the cover page of each of the code charts (<http://www.unicode.org/charts/>) that the Unicode Consortium provides:

“The shapes of the reference glyphs used in these code charts are not prescriptive. Considerable variation is to be expected in actual fonts.”

Disclaimer aside, how can we, merely looking at a code chart, determine the advance width, bearings, and other metrics for a glyph? The charts don’t include marks or guidelines for metrics. The code charts from the Unicode Consortium are a good starting point for and a lousy ending point for the development of a font family.
Comment 9 Etan Wexler 2008-06-29 12:33:05 UTC
In reply to comment #1, I declare that a quotation dash (U+2015 horizontal bar) whose length is the length of the quotation dash’s respective em dash (U+2014) is a bug. The same‐length situation is, at least, a bug in a scalable font for general use. (I feel generous enough to make an exception for monospace bitmap fonts whose design size is just a few pixels.) A quotation dash should be longer than its respective em dash.
Comment 10 Etan Wexler 2008-06-29 13:22:52 UTC
Comment #2 is incorrect on several counts. (I’ve edited the passages that I’ve quoted in order to use the intended punctuation and so forth.)

> So for example, Horizontal bar would be used like this:
> 
> ―“To be, or not to be.”

The horizontal bar (quotation dash) fits in the capacity of introducing quoted text, but the use of a quotation dash and the use of quotation marks are mutually exclusive at least in good American English typography and in good British English typography. For a given quotation, use one or the other, not both.

> Em dash would be used like this:
> 
> “To be, or not to be.” ——William Shakespeare
> 
> Note how em dash is usually used in pairs, and there is supposed to be a small
> gap between the two dashes.

Please read section the subsection “Dashes and Hyphens” of section 6.2 of version 5.0 of the Unicode Standard (<http://www.unicode.org/versions/Unicode5.0.0/ch06.pdf>):

“U+2014 em dash is used to make a break—like this—in the flow of a sentence. (Some typographers prefer to use U+2013 en dash set off with spaces – like this – to make the same kind of break.)”

It’s true that a comment in the code chart for the General Punctuation block (<http://www.unicode.org/charts/PDF/U2000.pdf>) mentions use in pairs. Given the example in the prose, it is clear that the pairs of em dashes are like pairs of parentheses, pairs of quotation marks, or pairs of vertical lines: one punctuation mark introduces a phrase or a quotation or the notation of a value, and the second punctuation mark terminates said phrase or quotation or notation of a value. More importantly, the foregoing is clear given centuries of typographic tradition in daily practice.

I am curious about the origin of the idea that a small space should appear by default between consecutive em dashes. As far as I know, there should be no space. I find that, in print, em dashes typically have zero bearings or tiny bearings on both sides. It would be typical, therefore, to see an em dash touching a preceding lowercase letter “e” or “o”. As for DejaVu fonts, having zero bearings on both sides of the em dash seems like a reasonable compromise because the bearings on letters and figures provide a small space in addition to or even despite the bearings of the em dash.
Comment 11 Etan Wexler 2008-06-29 13:48:37 UTC
Comment #4 is incorrect on several counts:

> Also, EN dash is supposed to be 1/2 the size of EM dash (as far as I understand
> it).

This is not necessarily the case. The names of the em dash and en dash reflect the origins of their designs. An em dash, originally, was as long as a capital letter “M” was wide. An en dash, originally, was as long as a capital letter “N” was wide. Of course, designs and usage have evolved. Still, I’ve never heard of a typographic rule that the length of an em dash should be twice the length of its respective en dash.

> As far as I can see, en dash is slightly larger than hyphen, and em
> dash is twice as big as en dash (but still small enough to have a small gap
> between two em dash characters). And horizontal line is the largest and has no
> gap between characters.

A en dash should be distinctly longer than, not slightly longer than, a hyphen. (A hyphen may be thicker than its respective dashes. A hyphen may be tilted from lower left to upper right.) Interglyphic gaps are possible regardless of the dash length. We have left and right bearings for this purpose.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.