David (the gdtoa author) got back to me and confirmed this is an upstream bug, and that my fix attached previously was the right one. He is trying to get that fix uploaded to the right ftp server soon. For future mingw-w64 reference, my change-controlled copy of his current tarball is updated here for future reference: https://github.com/vtjnash/gdtoa/tree/91156f01f80649aa5657b55bddffc8f0ad447814
On Thu, May 11, 2023 at 11:34 AM Jameson Nash <vtjn...@gmail.com> wrote: > No, I missed your message since I am not on the mailing list itself. > Thanks! > > Whelp, you apparently found a bug introduced into gdtoa itself, since the > same test program run on linux against their current code also fails with > the same incorrect byte ordering of the output: > > gdtoa$ ./a.out > > got: 1.06074e-314 > > bits: 000000007ff80000 > > I am not exactly sure of the intent here (since gdtoa doesn't use any sort > of version control), so I have mostly just removed that line of the change. > I will send a message to the gdoa developer also, informing him of your > finding > > My test code is attached as a github comment on the same commit attached > as a patch here. > > https://github.com/vtjnash/mingw-w64/commit/561c763593981f7516ab57b0cae974ebdcef4de3#commitcomment-112904299 > _______________________________________________ Mingw-w64-public mailing list Mingw-w64-public@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mingw-w64-public
