soft hyphen and hyphenation

Post by Johannes Hofmann
is it correct that we use a soft hyphen (0xc2 0xad) to draw the
hyphen when breaking words? It seems to draw ok with fltk/X11, but
wouldn't an unconditional hyphen 0xE2 0x80 0x90 be more correct?
I'm asking because e.g the Apple drawing library doesn't seem to
draw the soft hyphen if it thinks there should be no line break.

I see no problem with using unconditional breaks, and it seems to work
well with fltk/X11, so I've pushed the changes to hg.dillo.org.

(BTW: unconditional hyphens should also added do dillo_hyphen.)

Sebastian

corvid

2012-11-26 22:55:04 UTC

I just wanted to mention that if you compile with -pedantic, it complains like

textblock.cc:941:22: warning: overflow in implicit constant conversion [-Woverflow]

about

text[p++] = 0xe2;
text[p++] = 0x80;
text[p++] = 0x90;

Sebastian Geerken

2012-11-27 17:22:34 UTC

Post by corvid
I just wanted to mention that if you compile with -pedantic,

I get 633 warnings when I compile with -pedantic. :-)

Post by corvid
it complains like
textblock.cc:941:22: warning: overflow in implicit constant conversion [-Woverflow]
about
text[p++] = 0xe2;
text[p++] = 0x80;
text[p++] = 0x90;

I don't know how to deal best with this, but following change seems to
work:

text[p++] = '\xe2';
text[p++] = '\x80';
text[p++] = '\x90';

Sebastian

corvid

2012-11-27 18:02:04 UTC

Post by corvid
I just wanted to mention that if you compile with -pedantic,

I get 633 warnings when I compile with -pedantic. :-)

Yeah, I went through maybe a month ago and fixed a few assorted ones,
but the rest fell into a couple of categories that seemed best left
alone...

Johannes Hofmann

2012-11-27 19:43:37 UTC

Post by corvid
I just wanted to mention that if you compile with -pedantic,

I get 633 warnings when I compile with -pedantic. :-)

Post by corvid
it complains like
textblock.cc:941:22: warning: overflow in implicit constant conversion [-Woverflow]
about
text[p++] = 0xe2;
text[p++] = 0x80;
text[p++] = 0x90;

I don't know how to deal best with this, but following change seems to
text[p++] = '\xe2';
text[p++] = '\x80';
text[p++] = '\x90';

I know, I brought this up in the first place, but now I noticed,
that the unconditional hyphen (0xe2 0x80 0x90) seems not be
available in all fonts (e.g. I only see squares for it on
heise.de). So maybe a hyphen-minus aka '-' might be simpler and
commonly available (also see [1]). What do you think?

Cheers and sorry for the confusion,
Johannes

[1] http://www.cs.tut.fi/~jkorpela/dashes.html

corvid

2012-11-27 20:16:30 UTC

Post by Johannes Hofmann
I know, I brought this up in the first place, but now I noticed,
that the unconditional hyphen (0xe2 0x80 0x90) seems not be
available in all fonts (e.g. I only see squares for it on
heise.de). So maybe a hyphen-minus aka '-' might be simpler and
commonly available (also see [1]). What do you think?

hyphen-minus sounds safest to me.

I wish we had a good way to deal with the squares...

Sebastian Geerken

2012-11-27 21:16:36 UTC

hyphen-minus sounds safest to me.

What about making it configurable, with hyphen-minus as default?

Post by corvid
I wish we had a good way to deal with the squares...

One could think of testing which character is available, and so switch
to other characters, but probably this is unnecessarily complex.

Sebastian

corvid

2012-11-28 04:11:02 UTC

hyphen-minus sounds safest to me.

What about making it configurable, with hyphen-minus as default?

In general, it feels a little unusual to have so much configuration
related to hyphenation specifically, but...

Post by corvid
I wish we had a good way to deal with the squares...

One could think of testing which character is available, and so switch
to other characters, but probably this is unnecessarily complex.

Is there a way to do that?

Sebastian Geerken

2012-11-28 11:09:52 UTC

hyphen-minus sounds safest to me.

What about making it configurable, with hyphen-minus as default?

In general, it feels a little unusual to have so much configuration
related to hyphenation specifically, but...

See my latest change. Could be made configurable in a simple way.