Invalid breaks in words with accents

[ogmios-voice]

Converting the following asciidoc document to pdf (using asciidoctor-pdf-1.5.0.alpha.17.dev w/ prawn-2.2.2)

:pdf-page-size: a5

= test

aàbècìdòeùf gáhéiíjókúlým nâoêpîqôrûs tñuõvãw xäyëzïaöbücÿd
aabbccddeef gghhiijjkkllm nnooppqqrrs ttuuvvw xxyyzzaabbccd

results in an invalid break in the first row. The resulting block of the pdf looks like this (for me):

aàbècìdòeùf gáhéiíjókúlým nâoêpîqôrûs tñuõvãw xäyëzïa
öbüc d
aabbccddeef gghhiijjkkllm nnooppqqrrs ttuuvvw
xxyyzzaabbccd

(Note: the ÿ character is invisible in the result, but that is a font issue.)

But the breaks should be consistent, while the current position of the break depends on whether a given letter is accented or not. The correct output should be:

aàbècìdòeùf gáhéiíjókúlým nâoêpîqôrûs tñuõvãw
xäyëzïaöbücÿd
aabbccddeef gghhiijjkkllm nnooppqqrrs ttuuvvw
xxyyzzaabbccd

See also: asciidoctor/asciidoctor-pdf#979 (comment)

[gettalong]

Can you provide a simple Ruby script (no dependencies except Prawn) to reproduce the problem?

[ogmios-voice]

Unfortunately I can't. I don't have too much ruby experience, no experience at all with Prawn, and I will lack any free time in the following period.

[gettalong]

No problem! Just note that the issue resolution may take longer without such a script.

[pointlessone]

I'm closing it in favour of #1295. I don't think fixing just accented characters would be useful or even possible without addressing the bigger issue.