This is a nightmare. PDF export just mangles all the text - well not all of it, just most of it. Randomly it leaves bits ok even with exactly the same font, styles, colours, weights, and the same master objects applied - you can see in the screenshots one spread in the doc is ok and one is not (most aren;t) . Some headings are ok but body text is mashed. There's no logic to this, nothing consistent about why and how it fails.
Publisher came out just in time for me to start these two jobs with a hard print deadline. After 2 - 3 days of arsing about trying to fix this problem after setting 120 pages, I thought I'd finally solved this by stopping using my FontBase font manager, removing all the Montserrat font versions and reinstalling them to system fonts and trying again for the upteenth time. So I started the other doc. But it's back again. The problem is, every time I do this reinstalling fonts process, the document seems to interpret the fonts as new fonts, substituting the old, then I have to go through and set everything again. Like hundreds of Latin names in italic in the body copy I will have to go through and manually do again and again, every time it happens. And many more issues besides will need to be addressed again.
This is a real problem. I see someone else has it to - I commented on that thread days back but there's no response on there. This is Windows 10 Pro, Montserrat (Google font) - marked as fully embeddable. I tried turning off the subset option on exports dozens of times and it makes no difference. It's extended this job into days more than it should have been.
I really like what you're trying to do with this suite of software but please please please get this sorted - people are trying to produce pro documents for print but we're falling at the last hurdle. It's pretty crashy too but the recovery so far has been reasonably good. Plenty of other niggles with things but the PDF export thing really really needs sorting though. Please can somebody look into this urgently?