#3653 Unexpected soft wrap not only at whitespaces (ASIAN LANG

[tvojeho]

Hi, the new patch (More soft wraps not only at whitespaces (ASIAN LANGUAGES) - ID: 2483695) caused some unexpected ways of soft wrapping the text (see attached pic).

When using fancy quotes (html char “) at the end of the line, it can be put on the next line alone, but if I use simple quote ("), the wrap function keeps it stuck to the previous word, as it should be.

Regards, tvojeho

Using jEdit version: 5.0pre1
Daily build: 2012-02-14_12-03-57
Java runtime version: 1.7.0_02-b13
OS name: Windows 7
OS version: 6.1
OS arch: amd64

[tvojeho]

[Max Funk]

Interesting example :-) I tried out.

Probably, the solution is: &ldquo is "left double quotation mark". I assume, "left" means, left of a citation. In this case, the line break would be left of the quotation mark. I tried out and found, &rdquo, "right double quotation mark" works fine. The simple "quotation mark" can be left and right, and therefore, a line break between the quotation mark and a word will not happen.

So, no bug?

[tvojeho]

Yes, the example was from a book I am currently reading ;)

I think that the custom of right/left quotation mark depends on the specific locale, even if originally &ldquo probably was meant as left mark. e.g. my locale is czech and LibreOffice Writer automatically converts quotation marks to fancy codes, using the pattern '„quoted text.“'.

So it may not be a bug as such, but there should at least be some way to customize this feature.

tvojeho

[Max Funk]

Then, the bug would be in LibreOffice converter...
These quotes are characters named "left" or "right", except „ that is the html code of "double low 9 quotation mark". That one is not named "Left", because there is no mirrored "right" equivalent (closing is with right double quotation mark).
I think, the weak point for the usage is, that left and right quotes look so similar; e.g. with round brackets, one would not come on the idea to write )this( instead of (this).

[Max Funk]

Btw, I tried out your example in BabelMap, a full-fledged unicode text editor (free, at babelstone.co.uk), and that one falls really into your trap, even with "right double quotation mark"...

[tvojeho]

So, maybe this issue does not belong in the bugs as I thought but in the features section, to be able to custom define non-whitespace marks that do not break line.

I think the different quotation mark usage practices, and perhaps even other marks, in various locales need more flexibility to accomodate different customs.

To devels: should I close this tracker and open a feature request?

tvojeho

[Max Funk]

I am sorry. My last comment was not correct.
I looked at http://en.wikipedia.org/wiki/Guillemets.

It is not like (right), )false(.
It is mostly like this: <<French>>, <<Greek>>, <<Italian>>...
but also: >>Danish<<, >>Finnish>>

And that will not be possible without setting the locale
of the buffer and passing it to the wrap function.

[tvojeho]

Closing and creating a feature request tracker

[Kazutoshi Satoda]

Analyzed in a derived FR #3489523, and the cause seems to be in the
implementation of BreakIterator (in JRE).
http://sourceforge.net/support/tracker.php?aid=3489523

So I put hard coded workarounds in r21221. Here reported problem can be
said to be fixed in that revision.