On Sat, 2015 Sep 26 23:59-0700, Paul Eggert wrote: > > Given all the problems mentioned (including some in the proposed > patches), let's give up on trying to support any such folks. If they > want to build gnulib-using software on z/OS, they'll have to build > with the default configuration in which char is unsigned. It wouldn't > be practical for us to try to support char being signed when standard > chars have the top bit set.
I wasn't quite sure where to draw the "not worth the trouble" line, but I think this is defensible. > With that in mind I installed the attached further patch, which > simplifies the recent changes to c-ctype quite a bit. It's a shame; I was impressed by the work you had done to support the signed chars. But in any event, test-c-ctype in Git d2de2a91 passes in an unsigned-char EBCDIC build on z/OS, and that will hopefully remain the case for a good long while. --Daniel -- Daniel Richard G. || sk...@iskunk.org My ASCII-art .sig got a bad case of Times New Roman.
