On Tue, Oct 25, 2022 at 11:16:40AM +0200, Hilmar Preuße wrote: > Control: tags -1 + help > > Am 24.10.2022 um 14:57 teilte Jonas Smedegaard mit: > > Hi, > > > The dvisvgm package fails to build from source when linking against > > Ghostscript 10: > > > > libtool: link: g++ -Wall -Wnon-virtual-dtor -I../libs/clipper > > -I../libs/variant/include -I/usr/include/freetype2 -I/usr/include/libpng16 > > -g -O2 -ffile-prefix-map=/build/dvisvgm-2.14=. -fstack-protector-strong > > -Wformat -Werror=format-security -Wno-mismatched-tags -Wl,-z -Wl,relro -o > > dvisvgm dvisvgm.o ./.libs/libdvisvgm.a ../libs/clipper/libclipper.a > > -lfreetype ../libs/ff-woff/libfontforge.a -lwoff2enc -lbrotlienc -lcrypto > > -lz -lxxhash -lpotrace -lgs -lkpathsea > > /usr/bin/ld: ./.libs/libdvisvgm.a(Ghostscript.o): in function > > `Ghostscript::error_name(int)': > > ./src/Ghostscript.cpp:382: undefined reference to `gs_error_names' > > > > > It seems to me that dvisvgm attempts to extract private information > > from the libgs library, which is no longer provided since release 10 > > of Ghostscript. > > > > Upstream has a commit for this: > > https://github.com/mgieseki/dvisvgm/commit/9bf81fd0b6e7876e5079e917ed7e12163b9e7f7f > > The commit message is: "dropped usage of gs_error_names() because it's no > longer accessible as of GS 10.0.0" > > I can confirm that dvisvgm builds again, after applying the patch. I'm not a > C programmer, but the patch looks harmless to me. Could anybody confirm?
In general, trusting what upstream has already applied is a good default (and you will usually anyway get this in the next upstream version). > Hilmar cu Adrian
