I'll take that as a bug report...
-- Jon
On 07/24/2013 05:50 PM, Martin Buchholz wrote:
You're not the only one annoyed by the giant sample code font. I
agree with you that the "regular" text should be regular text size,
and the sample text not too much different - that's why we have
different fonts! CSS owners, please fix!
On Mon, Jul 22, 2013 at 4:14 PM, Mike Duigou <[email protected]
<mailto:[email protected]>> wrote:
I have been working on javadoc for the JDK 8 lambda streams
feature and ran across some interesting quirks in the current
default javadoc css.
We have been using the construction
<pre>{@code
...
}</pre>
for block examples. We add the {@code} to the normal <pre> block
to avoid having to use html entities within the sample to escape
the "&" and "<" characters. This makes the examples easier to read
in the original source file.
The formatting of a <pre> block and a <pre>{@code block is
slightly different as a result relative sizing and nesting.
The <pre> tag sets the font-size to 1.3em (stylesheet.css, line 31)
The <code> tag (emitted by {@code}) sets the font-size to 1.2em
(stylesheet.css, line 55)
When nested the effective size is default * body (76%) * 1.3 (pre)
* 1.2 (code).
It would be nice if {@code} nested inside <pre> didn't increase
the size. Using relative sizes is generally going to be weird
whenever nesting occurs especially if it can occur in more than
one order.
Could <pre> and <code> be made to use the same size?
Might it be better to use "<pre>{@literal ... " than "<pre>{@code
..." as {@literal doesn't add any styles?
Out of curiosity:
- Why is the default body size 76% of the default text size? The
100% size is supposed the user's comfortable reading size. Other
than for "fine print" why would we want to force a size smaller
than that?
- An explicit font selection is made for body copy but none is
made for code/pre text. Why not? Choosing a code font would allow
better matching of the size of the body copy and mono space text.
It would appear that for Arial/Courier that a 1.05em ratio is
somewhat better than the 1.2em currently used.
Thanks,
Mike
