Doing it for appearances, Hong Kong style!

by Michael S. Kaplan, published on 2011/01/13 07:01 -05:00, original URI: http://blogs.msdn.com/b/michkap/archive/2011/01/13/10115302.aspx

HKSCS, the Hong Kong Supplemental Character Set, is (if you wade through the extensive amount of text written on it, everywhere) can reasonably be purported as representing many different things:

Now the first two are what is most often discussed, while the third is usually taken for granted and not seen much except in a few obscure typography fora where people work to create such fonts.

Although nascently interested in the third point, I had much more expertise in the first two, and most of the work I had assigned to and the problems people were asking me about kept dragging me in on the first two points.

All of this conspired with my lack of typographic expertise to keep me from ever delving into what is really the most interesting point for me, in the long run.

Well, there is a lot behind the four fonts that make up the Hong Kong Supplemental Character Set vs. not support on Windows:

Well, because although you can explicitly choose the font you want to get particular glyphs, you could also specify nothing and let the default system locale set to Traditional Chinese (Hong Kong) vs. Traditional Chinese (Taiwan) do the selection for you.

In this way, a very easy configuratoion setting could have a huge difference in overall look and feel.

Well, wait. Without knowing the degree or amount of difference, that could be overstatement.

I'll start with the various Microsoft Typography command line tools like BreakTTC.exe (needed to break up the mingliu.ttc file containing so many of these fonts) and FontDiff.exe (needed to look into some of the differences).

When you compare the non-HKSCS fonts to the HKSCS ones in FontDiff.exe (click on the images to see them at a larger size):

Now one of the things that was done here for backwards compatibility with that first issue I mentioned was adding many Han to the PUA in the HKSCS font. You cansee this in FontDiff.exe's results as well if you scroll to the PUA:

This is the only place in either font that you see a difference of no glyph vs. a glyph, and it knocks that 7906 number down to 3020.

And of course all of the 4886 Han in the PUA portion of the MingLiU_HKSCS font are in their proper code point position in MingLiU-ExtB_HKSCS, and some percentage of those Han may also be different from their MingLiU-ExtB brethren, but unless someone made a mistake in the one of the fonts those differences should be captured in the MingLiU-ExtB vs. MingLiU-ExtB_HKSCS comparison. The FontDiff.exe tool is not useful for making those comparisons but I am sure the MST folks have other tools to verify that sort of thing!

Okay, now let's take a quick gander at one of these differences. Looking at this one (double-click to see it bigger):

U+9f96 will work for us. My first thought was the glyf table, but that didn't work:

because the tool is considering them to be equal even when they look kind of different. so let's try something else with U+9f96.

Let's do the comparison by looking at glyphid 28478 in one font versus glyphid 34003 in the other, directly:

You can also see the difference visually without that middle comparison frame in Word:

Now I have been told by some people I know both on the "typography" side and the "living in Hong Kong working with the HKSCS standard fonts" side that of the 4720 Han, some differences are more marked than this example here and some are less. But in the long run they do give a different experience for a user in Hong Kong.

Of course in Windows for a long time there were bugs that happened insetting the default system locale to Hong Kong instead of Taiwan, and either for that reason (or for other reasons less the fault of Microsoft!), users may prefer the Taiwan style glyphs for these Han. And they can use the same original workaround to get this --- a Taiwan Default System Locale!

You can dig more info differences here yourself using these same tools I was using, if you are interested....

Now FontDiff.exe can't do much here, it is a TrueType tool created before OpenType even existed.

Now if you look at U+9f96 in Microsoft JhengHei on a system with either a US default system locale or a Hong Kong one makes no difference:

It is possible that Microsoft JhengHei solves this issue using a more OpenType-like solution, but that kind of support is something that Notepad/Wordpad/Word/Windows UI don't know about and definitely can lead to a less than optimal experience for people who want the HKSCS-preferred glyphs.

I really do hope I am wrong about this last bit and that perhaps other more conventional Han work correctly here. But I'll need a scosh of typography expertise to say for sure....

As a HKSCS user for a long time, I have to say that I hadn't noticed these difference so much.

If you have 20+ DynaDont (3rd party font maker specialized in making Chinese fonts) font installed, you'll see that each fonts have subtle difference in preference to express the strokes.

Take horizontal stroke for example, some will prefer a plain horizontal line, while other will prefer it to raise from left to right at a small degree. And for dot stroke with horizontal stroke below, some will just leave it as a dot, and others will convert it to vertical stroke with big rounded head, connecting the horizontal stroke below.

I agree that the glyphs should be consistant for the same font name, but perheps it should be forgivable because the HKSCS version exist later - after the Hong Kong government decided to standardize HKSCS.

As for whether this will lead to experience less than optimal here, I think I can say that we don't care that much because we've got used to it. (Say, we have no problem seeing "裡" and "裏", "匯" and "滙" as the same character because we suffered the same problem in newspaper publishing. But "邨" and "村" does matter because although they refer to similar concept, the context they used is a bit different so most Hong Kong resident will consider them as different word, yet most will not have problem to understand by read it a second time.)