Hi Christoph, On Mon, Feb 23, 2026 at 12:14:54PM +0100, Christoph Berg wrote:
So in short, please add README.Debian file for this. Maybe explaining "when should you use one over the other flavor" is out of scope, but at least the existence of the flavors should be mentioned.
I’ve done that in the latest upload, thanks for asking for it!
Found while packaging documentdb which uses the library. There I hardcoded -lbidgcc000 now, but I guess that wasn't the optimal choice. ppc64el already complained about it:
-lbidgcc000 seems like the right choice, based on the flags in https://github.com/documentdb/documentdb/blob/cd47c92ee84be9906116c615d1d4985eb3f7ba04/scripts/install_setup_intel_decimal_math_lib.sh#L69
/usr/bin/ld: /usr/lib/powerpc64le-linux-gnu/libbidgcc000.a(bid_binarydecimal.o) uses IBM long double, src/query/bson_compare.o uses IEEE long double
This is https://lists.debian.org/debian-release/2025/09/msg00266.html — intelrdfpmath has never been rebuilt in Debian since the compiler change. The new upload should fix that. Regards, Stephen
signature.asc
Description: PGP signature
