Open alerque opened 10 months ago
By the way don't worry about this sample being a veritable festival of hyphenation. It does occur naturally in Turkish prose, but the copy and metrics used here were engineered as a torture test of needing to hyphenate at awkward places. Having more lines hyphenated than not is expected in this situation.
Apparently different years of the Turkish Language Institute's guidance treat this differently, as to different publisher's style guides. It isn't clear whether any of the differing years or publishers did so as an accommodation to what they could accomplish with existing tooling or not. This current page for example is recommending dropping the hyphen instead of the apostrophe in these cases.
Turkish has an interesting feature where apostrophes inside words (which are very common) are valid break points, but the correct way to break at them is to remove them and replace them with the hyphen.
For cross references here is discussion on the same issue raised a few years back and solved in SILE.
Here is the torture test case adapted from SILE's
test/bug-355
:This sample in Typst v0.9 has 6 bad breaks: 3 trailing apostrophes and 3 leading. Here is the view if you don't happen to have the font to get the metrics off hand:
Here is source for the same page in SILE:
And here is what it looks like showing at least 4 valid cases of apostrophe replacements:
This occurred to be because of #2579 and affects the mechanics of how word tokenizing vs. hyphenation points vs. how they are actually output needs to happen internally.