0.7.19 and 0.7.20 will be about the same level of messy. The only difference with pdf conversion in 0.7.20 is that it takes each hyphenated word and does a simple analysis. If the word exists in the document without a hyphen then the hyphen is removed. If it can't find a match with a hyphen then the hyphen is retained.
0.7.19 always retains the hyphens, but otherwise formatting should be identical.
The bug is that currently the search for versions of the word without hyphens can match special characters which cause that failure.