On 9/10/07, Brion Vibber brion@wikimedia.org wrote:
simetrical@svn.wikimedia.org wrote:
--- trunk/phase3/skins/monobook/main.css 2007-09-09 23:58:10 UTC (rev 25705) +++ trunk/phase3/skins/monobook/main.css 2007-09-10 00:45:32 UTC (rev 25706) @@ -114,28 +114,22 @@ h1, h2, h3, h4, h5, h6 { color: black; background: none;
font-weight: normal; margin: 0; padding-top: .5em; padding-bottom: .17em;
font-size: 100%;+}
[snip]
+h1 .mw-headline { font-size: 188%; }
This kind of creeps me out, forcing all headings to have another container to bump the size back out.
They've had the extra container since I added a wrapper to align the edit link properly. You have a wrapper around the actual heading, a wrapper around the section edit link, and a wrapper around both together.
It also reduces the spacing, such as at the top of the page, which feels rather cramped.
That part kind of sucks. It's frankly a pretty good argument for pixel measurements. To restore old behavior, it would be necessary to manually multiply out the desired figures and set separate paddings/margins for each heading. Or, we could just make them all have the same padding, which should be like 1.5 times current.
There doesn't seem to be any way to get Opera and Firefox to both display the edit sections in consistent sizes using the inherit-and-reverse method, at least without browser-specific fixes. Firefox intelligently keeps pixel sizes to reasonable precision when inheriting, whereas Opera appears to round them down to the nearest pixel internally on every inheritance.
For instance, 12px text at 80% (h6) works out to 9.6 pixels. Opera's Developer Console displays the relevant text as being 9 pixels, and sure enough, the edit section link needs to be jacked up to 134% for it to display at 12px. Sure enough, 9px*134% = 12.06, while 9px*133% = only 11.97. The correct figure is of course 125%, because 125%*80% = 5/4 * 4/5 = 1, and this is how Firefox and IE appear to treat it. Maybe a bad interpretation of the provision that computed values are what are inherited in CSS?
Anyway, I'll just do an Opera-specific fix, since this change is mucking stuff up. But it would be exceedingly handy if we could use the CSS rem unit!