https://bz.apache.org/ooo/show_bug.cgi?id=125431
--- Comment #73 from [email protected] --- (In reply to Arrigo Marchiori from comment #72) > (In reply to damjan from comment #71) > > (In reply to Arrigo Marchiori from comment #70) > > > Hello, > > > > > > (In reply to damjan from comment #58) > > > > Created attachment 84969 [details] > > > > Implement digest functions using openssl > > > > > > FWIW, I encountered this problem under FreeBSD, using a AOO 4.1.2 build > > > from > > > the ports tree. > > > The patch applied correctly to the sources and solved the problem for me. > > > > > > Thank you, Damjan! > > > > > > FYI, I submitted your patch to the FreeBSD Bugzilla: > > > https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=206234 > > > > No problem, thank you Arrigo for testing and reporting back! > > > > Are you absolutely certain it's my patch that fixed it, and not something > > else? Can you please recompile AOO from scratch, exactly like you did to get > > it to work, but WITHOUT my patch, and verify the problem still happens? > > Just verified. Without your patch, password-protected files cannot be > created nor opened. With your patch, everything works smoothly. > > FYI, nss version on my system is 3.21.0 (also installed from the FreeBSD > ports tree). Thank you, that helps a lot. Now I know for sure NSS is failing somewhere. If it's something in your Mozilla profile that's breaking it, you might be able to get it to work without my patch by stopping all applications that are using NSS (usually in the form of libnss3.so) - these are all Mozilla applications like Firefox and Thunderbird, OpenOffice, LibreOffice and any others. Then move or rename your ~/.mozilla directory (don't delete - it contains your Firefox bookmarks and settings!). Then start OpenOffice and try again. If it works, you could mix files between the old and new directories and eventually narrow it down to the specific file that causes the problem. Maybe we could eventually learn something from that file... The thing is, we shouldn't be using a library that fragile, that commonly breaks in the field with disastrous consequences, even for something simple like calculating cryptographic digest functions, and that we have no easy way to debug or fix. Maybe once we better understand what's wrong, and can reproduce the problem and test, we can use NSS with different settings so its more resilient, but its license doesn't really suit us either, so an alternative would be welcome. I've started an email thread on our [email protected] list to discuss possibilities. -- You are receiving this mail because: You are the assignee for the issue.
