Found it this morning when my sound stopped working again, and with Afarr's help found this error during compile:
building 'xpra.x11.bindings.ximage' extension gcc -pthread -Wno-unused-result -Wsign-compare -DDYNAMIC_ANNOTATIONS_ENABLED=1 -DNDEBUG -O2 -g -pipe -Wall -Werror=format-security -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector-strong --param=ssp-buffer-size=4 -grecord-gcc-switches -specs=/usr/lib/rpm/redhat/redhat-hardened-cc1 -m64 -mtune=generic -D_GNU_SOURCE -fPIC -fwrapv -fPIC -I/usr/include/python3.5m -c xpra/x11/bindings/ximage.c -o build/temp.linux-x86_64-3.5/xpra/x11/bindings/ximage.o -Wall -Werror -fPIC xpra/x11/bindings/ximage.c: In function ‘__pyx_pf_4xpra_3x11_8bindings_6ximage_11XShmWrapper_8read_palette’: xpra/x11/bindings/ximage.c:8800:35: error: comparison between signed and unsigned integer expressions [-Werror=sign-compare] for (__pyx_t_10 = 0; __pyx_t_10 < __pyx_t_15; __pyx_t_10+=1) { ^
Reverting the server to r15122 builds fine with no errors, anything past r15123 will fail the Python3 build. Oddly enough the Python2 builds fine - I build 3 then 2 in my script and it'll still run afterwards. Weird.
Anyways, the change to x11/bindings/ximage.pyx
is the culprit here.
As always, using my Fedora 25 Server and Workstation machines doing a trunk build.
(tiny) fix in r15130
this ticket has been moved to: https://github.com/Xpra-org/xpra/issues/1448