The argument relies on 'visually different' being a well-defined, binary distinction. It is not. 'Similarity' and 'dissimilarity' can at best be described for a narrow set of purposes with guiding rules. So is the exact shape of the vertical or vertically-slanted dot on top of 言 enough to trigger the dissimilarity warning, or is it only when it's written as a horizontal? How about 吉田 vs. ⿱土口田? Is that similar or dissimilar? How about that little 'hair' stroke (ヒゲ or 筆押さえ) that is sometimes seen in characters like 文? When you compare the samples shown over at http://tonan.seesaa.net/article/431481813.html, wouldn't you agree that it is quite difficult to predict which characters should and which ones shouldn't have a ヒゲ? Did you know that the JIS 1983 edition did include the ヒゲ in many characters where the JIS 1990 edition omitted it?
You'll find all the gory details of what changed between JIS encoding editions in this fine book: 大熊肇: 文字の骨組[1], esp. pp164—178 where all the fine details between different editions of the JIS encoding are listed. It's mind-boggling! It's almost as though the Japanese standards body itself has been having slightly different opinions about their own writing system over the decades. Now if you're intent on getting all the fine details just right for your print edition you can either apply a newer or older JIS encoding to your document, OR invent that file format you're talking about where you can switch encodings mid-way, OR access OpenType font features from your document somehow... mmmh... how to do that... mmmh...
Oh I know! Let's use Unicode and HTML and CSS and get full access to OpenType font features for any point in the document on a per-codepoint basis.
For those folks who really pine for the bad old days when all we had was US ASCII plus an insane number of other encodings, please do read this single Wikipedia article: https://en.wikipedia.org/wiki/ISO/IEC_2022 it will change your mind.
You'll find all the gory details of what changed between JIS encoding editions in this fine book: 大熊肇: 文字の骨組[1], esp. pp164—178 where all the fine details between different editions of the JIS encoding are listed. It's mind-boggling! It's almost as though the Japanese standards body itself has been having slightly different opinions about their own writing system over the decades. Now if you're intent on getting all the fine details just right for your print edition you can either apply a newer or older JIS encoding to your document, OR invent that file format you're talking about where you can switch encodings mid-way, OR access OpenType font features from your document somehow... mmmh... how to do that... mmmh...
Oh I know! Let's use Unicode and HTML and CSS and get full access to OpenType font features for any point in the document on a per-codepoint basis.
For those folks who really pine for the bad old days when all we had was US ASCII plus an insane number of other encodings, please do read this single Wikipedia article: https://en.wikipedia.org/wiki/ISO/IEC_2022 it will change your mind.
* [1] https://www.amazon.co.jp/%E6%96%87%E5%AD%97%E3%81%AE%E9%AA%A...