At 2022-10-07T07:14:35+0000, Werner LEMBERG wrote: > > >> This inconsistency of gs on my GNU/Linux is of no importance, but > >> it clearly shows that there is a bug while compiling groff: It uses > >> `gropdf` from my system instead of using the just compiled > >> `gropdf`! > > > > Are you sure? I cannot verify this hypothesis; whether I build > > inside the tree or out of it, I can't get the build to use any groff > > executables that aren't in the tree. > > Good. I no longer have this gs inconsistency, making it hard to check > for the problem again.
Well, darn. > OK. You might also try `strace` to check exactly what files are > opened in case such a problem occurs again. > > > So maybe this is the same issue John Gardner reported after all. > > > > https://savannah.gnu.org/bugs/index.php?62918 > > I don't think so because I compiled groff directly from the tarball. But if I'm right, the problem doesn't have to do with the groff tarball, but with on-system non-groff resources. Note the similarity between your failure report and John's. gropdf:contrib/mom/examples/typesetting.mom: fatal error: failed to open '/usr/local/Cellar/ghostscript/9.56.1/share/ghostscript/9.56.1/Resource/Font/C059-Roman' gropdf:src/utils/addftinfo/addftinfo.1: fatal error: failed to open '/usr/local/Cellar/ghostscript/9.56.1/share/ghostscript/9.56.1/Resource/Font/NimbusRoman-Italic' gropdf:contrib/mom/examples/typesetting.mom: fatal error: failed to open '/usr/share/ghostscript/9.52/Resource/Font/C059-Roman' gropdf:src/utils/addftinfo/addftinfo.1: fatal error: failed to open '/usr/share/ghostscript/9.52/Resource/Font/NimbusRoman-Italic' (How many people reading this can tell which was whose?) These are much more similar than different. The same fonts are failing to be found during attempts to generate the same documents. The difference is in the details of the location gropdf is looking for URW font files to embed. I notice that our configuration report does not disclose where we found the URW fonts. GNU Troff version 1.23.0.rc1.3112-d7140 ---------------------------------------------------------------------- installation directory prefix : /usr/local C++ compiler and options : g++ -Wall -Wextra -Wformat=2 -Wstringop-overflow=4 -Wshadow=global -Wredundant-decls -Wunused -Wunused-parameter -fsanitize=signed-integer-overflow -fsanitize-undefined-trap-on-error -fstack-protector-strong -fno-common -fstack-clash-protection -ftrapv -funsigned-char -fvar-tracking-assignments -ggdb -Og -fcheck-new -Wredundant-decls -std=c++98 -pedantic -Wformat=2 -D_FORTIFY_SOURCE=2 use libgroff's memory allocator : no C compiler and options : gcc -Wall -Wextra -Wformat=2 -Wstringop-overflow=4 -Wshadow=global -Wredundant-decls -Wunused -Wunused-parameter -fsanitize=signed-integer-overflow -fsanitize-undefined-trap-on-error -fstack-protector-strong -fno-common -fstack-clash-protection -ftrapv -funsigned-char -fvar-tracking-assignments -ggdb -Og -Wmissing-prototypes -Wold-style-definition -Wstrict-prototypes -Wold-style-declaration -Wformat=2 -D_FORTIFY_SOURCE=2 Perl interpreter version : 5.32.1 X11 support : enabled X11 app defaults directory : /usr/local/lib/X11/app-defaults 'groff -l' uses print spooler : lpr use URW fonts for PDF output : yes preconv can use uchardet library : yes can build groff.dvi, groff.pdf : yes tests can use poppler PDF tools : yes ---------------------------------------------------------------------- I don't have enough information to solve this, but I do have two ideas that I think may help flush it out. 1. Add the purportedly successful URW font location path to the configuration report. 2. Have gropdf report more context when croaking as shown above. I _assume_ the failure was due to an embedding failure. But it would be good to be sure. John did attach his entire build output, so we can see where the configure script _did_ find his URW fonts: checking for URW fonts in Type 1/PFB format... found in /usr/local/Cellar/ghostscript/9.56.1_1/share/ghostscript/fonts That was the failed build. The successful one? checking for URW fonts in Type 1/PFB format... found in /usr/local/Cellar/ghostscript/9.56.1_1/share/ghostscript/fonts Indeed. So our configure script is finding the correct location but gropdf _sometimes_ doesn't use it. In John's case, it looked in gropdf:contrib/mom/examples/typesetting.mom: fatal error: failed to open '/usr/local/Cellar/ghostscript/9.56.1/share/ghostscript/9.56.1/Resource/Font/C059-Roman' We may have to go to RC2 without a "fix", but maybe testers benefiting from the foregoing two items can supply enough information for us to kill it. Regards, Branden
signature.asc
Description: PGP signature
