Hello Branden, here is a complete list of issues that prevent me from building, FYI:
1. https://savannah.gnu.org/bugs/index.php?68716 For now, i'm working around that crash by simply ignoring failure, using this hack at two places in doc/doc.am: - $(tmac_srcdir)/en.tmac $(GROFF_MAN_PAGES3) > $@ + $(tmac_srcdir)/en.tmac $(GROFF_MAN_PAGES3) > $@ \ + || echo "$(DOC_PDFMOM) $@: exit status $$?" 2. texi2dvi returns non-zero exit status The OpenBSD-current port of OpenBSD provides: $ gtexi2dvi --version texi2dvi (GNU Texinfo 7.1) The non-zero exit status from that one aborts the build, i do notyet understand why it returns non-zero. It prints no error message, all it says is this: Output written on groff.dvi (317 pages, 976284 bytes). Transcript written on groff.log. /usr/local/bin/gtexi2dvi: etex exited with bad status, quitting. For now, i'm working around that crash by simply ignoring failure, using this hack at two places in doc/doc.am: $(PROG_TEXI2DVI) -e --batch --build-dir=doc/`basename $@`.t2d \ - -o $@ $< + -o $@ $< || echo "texi2dvi $@: exit status $$?" 3. In the DOC_LINE_LAYOUT_EPS rule, i need this: -$(DOC_LINE_LAYOUT_EPS): doc/line-layout.roff pic groff troff grops \ +$(DOC_LINE_LAYOUT_EPS): $(doc_srcdir)/line-layout.roff pic groff troff grops \ font/devps/stamp $(AM_V_GEN)$(MKDIR_P) `dirname $@` \ - && $(DOC_GROFF) -p -T ps -P -b16 $< >$(DOC_LL).ps \ + && $(DOC_GROFF) -p -T ps -P -b16 $(doc_srcdir)/line-layout.roff \ + > $(DOC_LL).ps \ 4. In the DOC_LINE_LAYOUT_PNG rule,i need this: - && convert -density 300 $< PNG24:$@ + && convert -density 300 $(DOC_LINE_LAYOUT_EPS) PNG24:$@ With these six changes, the build completes, but obviously not building everything completely. This will now allow me to start testing patches, but it did not start run-time testing yet. G. Branden Robinson wrote on Sat, Sep 26, 2026 at 10:26:45AM -0500: > There's one more GNU-ish use of `$<` in "doc/doc.am". > > At 2026-09-25T12:49:05-0500, G. Branden Robinson wrote: > > $ git blame doc/doc.am > > 931c31ab4d (G. Branden Robinson 2022-05-21 09:26:20 -0500 387) # Use '-K > utf8', not '-k', in case 'configure' didn't find uchardet. > 92349ae223 (G. Branden Robinson 2022-05-30 12:42:38 -0500 388) # The French > translation uses tbl; its English counterpart does not. > f8ccb7da6b (G. Branden Robinson 2023-09-22 01:27:57 -0500 389) > doc/meintro_fr.ps: doc/meintro_fr.me preconv tbl > bdd9c3605e (G. Branden Robinson 2021-12-05 12:41:37 +1100 390) > $(GROFF_V)$(MKDIR_P) `dirname $@` \ > 92349ae223 (G. Branden Robinson 2022-05-30 12:42:38 -0500 391) && > $(DOC_GROFF) -K utf8 -t -Tps -me -mfr $< >$@ > > I can't account for why neither of us hit this, given its age. I can, by citing from the OpenBSD make(1) manual page: For maximum compatibility, `<' should only be used for actual inference rules. It is also set for normal target rules when there is an inference rule that matches the current target and prerequisite in scope. That is, in .SUFFIXES: .c .o file.o: file.c cmd1 $< .c.o: cmd2 building file.o will execute "cmd1 file.c". Consequently, the doc/meintro_fr.ps rule works on OpenBSD because there is also a ".me.ps:" rule in doc/doc.am - but that that works is an extension and not portable to POSIX make. > A fix for that , along with the other extended uses of `$<`, will be in > my next push to the groff `master` branch (which will _not_ be tagged > 1.25.0.rc3 soon after and will not have the security fixes currently in > preparation). Sounds good. As soon as you push those fixes to master, i will use those instead of my own fixes 3 and 4 listed above. That leaves items 1 and 2 above on the list of crashes to fix. Yours, Ingo
