Typesetters of the world - ARISE !

[HarryT]

Originally Posted by tompe

Breaking a line at an emdash seems to me to be bad typography. It is something that stands out and will attract attention.

... but it's not nearly as bad as adding a hyphen to an emdash, as though the dash was an alphabetic character, is it? I would regard the Gen3's behaviour in doing that as a bug, plain and simple.

[tompe]

Originally Posted by HarryT

... but it's not nearly as bad as adding a hyphen to an emdash, as though the dash was an alphabetic character, is it? I would regard the Gen3's behaviour in doing that as a bug, plain and simple.

I agree. I assume you have reported that to Bookeen?

[HarryT]

Not yet, because I didn't know it would do anything so dreadful until Jellby told us a few posts ago! I certainly shall report it, now I know about it, and I'd encourage other people to do likewise. I'll try to create a simple file which illustrates the problem, and send it to them.

[Jellby]

Originally Posted by tompe

I think we have discussed it before here. TeX do not do it unless you force it to do it.

But it does break at hyphens, it just uses a different configuration (\exhyphenpenalty, if I remember correctly).

Originally Posted by tompe

In Swedish we use spaces around emdashes. In TeX I prefer to put a thin space before and after an emdash. Breaking a line at an emdash seems to me to be bad typography. It is something that stands out and will attract attention.

When typesetting a book that will be printed or "fixed" as PDF. There's many things you can try to tweak and avoid. Yes, avoiding breaks at explicit hyphens, dashes, even after the word "I", is desirable. But with a reflowable text in a small screen, you cannot be as demanding as with a printed book (yet). I prefer a break at a dash (which, by the way, I don't perceive as much distracting) over an underfull line or a paragraph full of hypenated breaks.

As you suggest, though, there is a problem with different typographic conventions in different languages. In Spanish, for instance, dashes are used like parentheses: with a space on the "outside" and no space on "inside". Breaking the line at the space is quite right, but breaking it between the dash and the inner word would be plain wrong. Fortunately, the current behaviour on the Cybook is fine for this. In English, I prefer to create my books with "space en-dash space" instead of "em-dash" (as I've seen done in some printed books too), that way I let the reader break the line at the dash and, at the same time the dash is a bit shorter... but I would prefer to have breaking zero-width spaces around an em-dash.

[tompe]

Originally Posted by Jellby

But it does break at hyphens, it just uses a different configuration (\exhyphenpenalty, if I remember correctly).

Maybe the penalty is just so high that I have not seen it very often and assumed it was a hard rule.

When typesetting a book that will be printed or "fixed" as PDF. There's many things you can try to tweak and avoid. Yes, avoiding breaks at explicit hyphens, dashes, even after the word "I", is desirable. But with a reflowable text in a small screen, you cannot be as demanding as with a printed book (yet).

I wonder if this is true. Maybe the TeX typesetting algorithm will work for ebook readers. As I see it the most important things is that you have to typeset on the paragraph level and not just line for line.

[Jellby]

Originally Posted by tompe

Maybe the penalty is just so high that I have not seen it very often and assumed it was a hard rule.

I think the default is the same as for any hyphenated word...

I wonder if this is true. Maybe the TeX typesetting algorithm will work for ebook readers. As I see it the most important things is that you have to typeset on the paragraph level and not just line for line.

That would, of course, be a great improvement. But there are two remaining "problems":

- If you are going to use a small screen or a large font, the lines are going to be shorter than what's usual in printed books.

- If top-quality typography is desired, one has sometimes to add some manual tweaks here and there, or even rewrite a passage (to avoid long, hyphenated words, to add a couple of lines and prevent a widow...). This is fine if you are going to typeset once and for all, but it's not feasible for a reflowable text.

Both these things mean, in my opinion, that for ebooks one has to allow for more flexibility and be a bit more permissive than with printed books. Still, I agree there's room for improvement and we should try to get the best possible result.

[tompe]

Originally Posted by Jellby

I think the default is the same as for any hyphenated word...

I just made a test file and you are right. I wonder if some default or functionality have changed recent versions of LaTeX. Or maybe it is just my memory that is broken

That would, of course, be a great improvement. But there are two remaining "problems":

- If you are going to use a small screen or a large font, the lines are going to be shorter than what's usual in printed books.

- If top-quality typography is desired, one has sometimes to add some manual tweaks here and there, or even rewrite a passage (to avoid long, hyphenated words, to add a couple of lines and prevent a widow...). This is fine if you are going to typeset once and for all, but it's not feasible for a reflowable text.

Both these things mean, in my opinion, that for ebooks one has to allow for more flexibility and be a bit more permissive than with printed books. Still, I agree there's room for improvement and we should try to get the best possible result.

I agree.

I think you have to accept a certain amount of underfull and overfull hboxes. But wouldn't it be fun to have counters for underfull and overfull hboxes displayed when reading the text...

[JSWolf]

The problem is that if you have something like word—word where the software takes it as one word where it ends up as the beginning of a line, the previous line usually will be sitting there with rather large spaces between the words. Would it not be better to break at em dashes only if it will prevent a line from having too much space between words?

[HarryT]

Originally Posted by JSWolf

The problem is that if you have something like word—word where the software takes it as one word where it ends up as the beginning of a line, the previous line usually will be sitting there with rather large spaces between the words. Would it not be better to break at em dashes only if it will prevent a line from having too much space between words?

I'd prefer it personally if it would regard the end of the dash as a natural break-point like a space.

What does the Sony Reader do when it comes to breaking lines on dashes?

[llasram]

Originally Posted by HarryT

What does the Sony Reader do when it comes to breaking lines on dashes?

The LRF renderer never breaks around an em dash. AdobeDE considers all em dashes fair game to break around. The former results in the bad breaks you'd expect, but the latter can result in things like:

“This is a line of dialog which ends with a dash
—”

Oops. Fortunately AdobeDE does respect CSS ‘white-space: nowrap’, which can be used to selectively prevent breaks, although to be honest I’ve only actually seen this used in EPUB books I’ve produced myself.

-Marshall

[TedPark]

Do you guys know if the newest firmware does any better on this? Do you know if there are any firmware updates coming down the pike soon?

And - since I've been out of the loop for several months - can you point me to the latest firmware available here from MobileRead? I also want Unified Font and the Improved Icons. Yes, I know, I can search - but if you already knew, it would save me some effort.