On Sun, Nov 11 2018, Daniel Dickman <didick...@gmail.com> wrote: > Update to gprolog 1.4.5 which allegedly fixes many bugs. Also drop > maintainer due to my inability to reach them for close to a decade. > > While here, eliminate use of sbrk. gprolog uses Doug Lea malloc 2.8.5 > which can be configured to use mmap instead of sbrk on platforms like > MacOS. So let's do the same on OpenBSD. > > NEWS says a lot of bugs were fixed, so it's also possible this helps to > build gprolog with gcc 6 (the problem naddy reported). Although I haven't > tested this. > > Some ideas after this goes in based on NEWS: > - sparc64 might be supported now, if someone tests it then it can be added > to ONLY_FOR_ARCHS
If it was fixed, it got broken again. The broken detection of a working sigaction leads to some code that dereferences a pointer to struct sigcontext, which is inherently unportable. I have a fix for this though. > - there's mention of lots of clang fixes, so we might be able to switch > compilers. > > Tested on i386. ok? make test also passes on amd64 (and sparc64 after the sigaction fix) ok jca@ -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
