Danda and wrapping

[r12a]

2.8.2 Punctuations https://w3c.github.io/iip/gap-analysis/beng-gap.html#danda

Before getting to the main comment, this would be a good time for me to get the answer to a couple of questions i have wanted answers to for a long time:

• Should there or can there be a space between a word and a following danda (or double danda)?
• Should ॥ [U+0965 DEVANAGARI DOUBLE DANDA] be used for the double danda, or can one use two of । [U+0964 DEVANAGARI DANDA]?
• Is a single space adequate after a danda and the first word of the next phrase?

Now to the main comment:

The properties of Danda and Double Danda should be the same as the properties of FullStop or other punctuation marks, and a new line should not begin with Danda and Double Danda.

It would be good to have empirical evidence here as to whether or not this is a problem. To that end, i created a test at https://w3c.github.io/iip/gap-analysis/bengali-tests/bengali-danda-001 (I can create additional tests once i have answers to the above questions.)

My findings on Mac High Sierra with latest browser versions are that Firefox, Safari, and Chrome are OK. On Windows 10, Edge is also OK. So there may not be an issue here. (Although the information is worth adding to the blreq document, once we start on that.)

[tiroj]

Our take on your first question at Tiro is that there should not be a space character between a word and a following danda, but instead the customary gap should be incorporated in the advance width of the danda and double danda glyphs. Note that there is a secondary use of dandas which is to set apart section or verse numbering, in which the number is placed between pairs of double dandas. To obtain the correct spacing, the character sequence is usually <double danda, space, numeral(s), double danda>.

With regard to the second question, we strongly encourage authors to use the double danda character and not a sequence of two single dandas. The practice of using a pair of single dandas can be traced to typewriters and hot metal typesetting machines that had a single danda key. If the single danda is correctly spaced, as described above, two individual dandas will be too loosely spaced. [I really don't want to get into e.g. kerning or ligating the two dandas to mimic a double danda glyph.]

I've found a single word space after the danda to be adequate.

Your statement re. dandas being treated as fullstop is correct. You never want a danda split off from the preceding text and wrapping to the beginning of a new line.

[lianghai]

Dandas are preferred to have symmetrical whitespace surrounding them, and such whitespace should be achieved by a pair of space characters. Having a preceding whitespace built-in in the danda glyph is a bad hack to prevent creating a preceding linebreak opportunity.

The proper way to disallow a linebreak opportunity in <word, space, danda> should be giving danda characters Line_Break = EX (see UAX #14), like what we’ve already done for Latin exclamation mark, question mark (both for the French style punctuation spacing), as well as Mongolian comma and period (these Mongolian ones are the very same case, also suffering from font producers making built-in whitespace).

[tiroj]

Dandas are preferred to have symmetrical whitespace surrounding them

That's not what our newspaper publisher and book design clients have preferred, and they format their text without a space character before the danda. I'm quite willing to believe that there are different preferences and practices, and a lack of standardisation in either editorial practice or how fonts are built.

I agree with @lianghai that the danda characters should have Line_Break = EX property to disallow a linebreak when a space is present before the danda, but that should be the case regardless of whether the space is encouraged or not.

[lianghai]

I'm quite willing to believe that there are different preferences and practices, and a lack of standardisation in either editorial practice or how fonts are built.

Yep, I agree with this.