Re: [css3-text] Comments on hyphenation

On 16 May 2011, at 13:56, Robert O'Callahan wrote:

> On Sun, May 15, 2011 at 6:16 AM, Mathias Nater <mathiasnater@gmail.com> wrote:
>> A missplaced hyphen is a spelling error and thus not acceptable on a website.

FWIW, I'd consider a misplaced hyphen to be a layout (or formatting, or typesetting) error rather than a spelling error, and I'd consider it somewhat less unacceptable than a spelling error - although still something to be avoided, of course.

>> Automatic hyphenation can never be perfect due to ambiguous words.

It can probably never be "perfect" in the sense of reliably identifying every possible hyphenation point; however, it should be possible for it to be "good enough", in the sense of identifying a large enough number of legitimate hyphenation points to substantially improve layout, while avoiding incorrect or ambiguous cases.

>>  Therefor a webdeveloper has to be able to check and fix hyphenation.
>> 
> Since hyphenation isn't standardized (and shouldn't be, IMHO), you can only check hyphenation in particular browsers, and other browsers might do something different. I think authors who need total control over hyphenation should hyphenate their own text using &shy; and hyphens:manual.

Yes, this is the only way to guarantee complete control (assuming browsers implement &shy; and hyphens:manual properly). There is no clear specification of the algorithms or resources to be used for hyphens:auto, and therefore there is no expectation of 100% consistent behavior. What _can_ reasonably be expected is that all browsers that support hyphens:auto should do a "good enough" job (as above), but they may not all do exactly the _same_ job (just as they may not all use exactly the same fonts and metrics, or apply the same glyph hinting and antialiasing, or implement exactly the same microtypographic features such as ligatures and kerning, etc.)

> I'm not too excited about implementing hyphenate-resource. It seems unlikely to me that a significant number of Web developers will bother developing and deploying their own hyphenation dictionaries.

Right; especially given the lack of a clear standard for the format of such resources. Even though most hyphenation algorithms in current use (AFAIK) are derivatives of Liang's work in TeX, there are some variations and extensions, and therefore the resources/dictionaries are not necessarily 100% interoperable.

JK

Received on Monday, 16 May 2011 14:53:08 UTC