Re: [NTG-context] A bug in latin modern fonts?

On Sat, 4 Feb 2012, Janne Junnila wrote:

On Sun, Jan 29, 2012 at 12:42 PM, Janne Junnila wrote:

...

Hi!

The following example seems to have some troubles with the latin modern fonts:

\starttext

The function $f \colon \reals \to \reals$ is given by $x \mapsto x - 3$.

\stoptext

The colon is printed fine, but the blackboard bold Rs don't look correct. The \mapsto symbol is missing and there's a hyphen instead of minus.

Are others having the same issue? I tried to do a fresh install, but the problem persisted. There were some messages like

fonts           > otf loading > skipping cyclic reference U+00028 in math variant U+00028

during compiling. I don't know if they are related to the problem.

Hi,

I would like to my previous message that also Asana-Math seems to have problems.

\setupbodyfont[asana] \starttext

The function $f \colon \reals \to \reals$ is given by $x \mapsto x - 3$. Its derivative is $f'(x) = 1$.

\stoptext

In the above the ' in f'(x) looks wrong and there's hyphen instead of minus at x - 3. The problems arise in the current standalone, the latest stable one worked OK at least for the latin modern fonts.

The reference implementatation for math handling is cambria. With cambria $f'(x)$ and $\mapsto$ work fine. But $a-b$ gives text minus, and $a \colonequals b$ gives an error. @Hans and Khaled: Any idea why this is happening? \starttext \setupbodyfont[cambria] $a-b$ % text minus $a \colonequals b$ % error \stoptext $f'(x)$ works fine with cambria and xits, but is placed low with asana. My guess is that this is the font's fault; with asana $f^{'}$ works fine. The same "bug" is also present with LMmath. The symbol \mapsto appears to be missing from asana and lmmath. ConTeXt currently does not use fall backs for missing math glyphs. Finally, $\reals$ works correctly with cambria, asana, lmmath, and xtis. It does not work with euler, but that is because euler is missing blackboard gylphs. So, to summarize: 1. - (minus), is clearly a bug in ConTeXt. 2. \colonequals is a bug in ConTeXt's definition. We should be using the glyph from the font anyways and falling back only if the glyph is missing. 3. \mapsto glyph is missing from lmmath and asana. We need to fall back to composed glyphs. 4. \reals is working fine. The shape of the glyph in lmmath is unusual, but that should be a font issue rather than a context issue. Aditya