Hi Abel, i committed a tweaked version of your patch with these changes:
- pass pointers to structs around, don't pass structs by value - no need for an additional automatic variable for the URI - no need for comments similar to: i++; /* increment i */ - only use file:// when -O tag was indeed given I dropped changing the filename. I consider that a bug in lynx(1). Changing behaviour based on the filename "extension" is Microsoft Windows style and a concept totally alien to Unix. So, i don't really like the idea to cater to that. Besides, it's trivial to work around the lynx bug with -force_html without putting weird, pointless complication into mandoc. Your patch was mangled in at least one way. Most lines were prefixed with a bogus space character; i did not check whether it was also mangled in other ways. In this case the mangling didn't hurt because i had to apply it by hand anyway to check it and to perform the above tweaks. But when you send patches in the future, please try to make sure they apply cleanly. Mangled patches can sometimes cause needless annoyance, and they often delay merging by OpenBSD developers. I have also documented the new feature in the "HTML Output" subsection of the mandoc(1) manual page. Thanks for the suggestion and for the patch! I is very rare that users who are not BSD or Linux operating system developers send such substantial mandoc patches. The latest instance i found in my notes came from Franco Fichtner in 2013. Yours, Ingo
