Paul Eggert <[EMAIL PROTECTED]> wrote: > Matt Kraai <[EMAIL PROTECTED]> writes: >> , which is still present in the latest Gnulib sources. time_t is >> equivalent to an unsigned long, which causes both of the comparisons >> to fail. > > It looks like someone has to port getdate.y to QNX 6, then. Do you > have the time to do that? QNX 6 is certainly entitled by POSIX to > have an unsigned long time_t, but that (ahem) unusual implementation > will break the code in 'getdate'. > > One hacky workaround is to remove the offending "verify" line. The > resulting code will fail miserably with some time stamps, but perhaps > you won't run into any of those time stamps.
To get an idea of what can go wrong in that case, try building coreutils ftp://alpha.gnu.org/gnu/coreutils/coreutils-6.9.92.tar.gz ftp://alpha.gnu.org/gnu/coreutils/coreutils-6.9.92.tar.lzma with it, and then run "make check". There should be tests for at least some of the corner cases in tests/misc/date.
