An port may use symbolic links for manual pages. That's fine. However, if some of these links are absolute symlinks we are in trouble. It will not work if you move the manual pages around and use them in other places, e.g. as on https://www.freebsd.org/cgi/man.cgi I checked the packages on pkg.freebsd.org/freebsd:12:x86:64/latest/All and found these 4 packages with absolute symlinks for man pages: ./libdca-0.0.5_1.txz ./gnuit-4.9.5.txz ./ncbi-toolkit-2012.06.20_1.txz ./xli-1.17.0_16.txz The absolute links needs to be replaced by relative symlinks or hard links. in detail: ./libdca-0.0.5_1.txz:lrwxr-xr-x 0 root wheel 0 Nov 16 01:39 /usr/local/man/man1/dtsdec.1.gz -> /usr/local/man/man1/dcadec.1.gz ./libdca-0.0.5_1.txz:lrwxr-xr-x 0 root wheel 0 Nov 16 01:39 /usr/local/man/man1/extract_dts.1.gz -> /usr/local/man/man1/extract_dca.1.gz ./gnuit-4.9.5.txz:lrwxr-xr-x 0 root wheel 0 Nov 17 19:21 /usr/local/man/man1/gitregrep.1.gz -> /usr/local/man/man1/gitrgrep.1.gz ./gnuit-4.9.5.txz:lrwxr-xr-x 0 root wheel 0 Nov 17 19:21 /usr/local/man/man1/gitrfgrep.1.gz -> /usr/local/man/man1/gitrgrep.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/bl2seq.1.gz -> /usr/local/man/man1/blast.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/blastall.1.gz -> /usr/local/man/man1/blast.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/blastcl3.1.gz -> /usr/local/man/man1/blast.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/blastpgp.1.gz -> /usr/local/man/man1/blast.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/impala.1.gz -> /usr/local/man/man1/blast.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/megablast.1.gz -> /usr/local/man/man1/blast.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/rpsblast.1.gz -> /usr/local/man/man1/blast.1.gz ./ncbi-toolkit-2012.06.20_1.txz:lrwxr-xr-x 0 root wheel 0 Dec 7 02:13 /usr/local/man/man1/seedtop.1.gz -> /usr/local/man/man1/blast.1.gz ./xli-1.17.0_16.txz:lrwxr-xr-x 0 root wheel 0 Nov 17 03:45 /usr/local/man/man1/xview.1.gz -> /usr/local/man/man1/xli.1.gz ./xli-1.17.0_16.txz:lrwxr-xr-x 0 root wheel 0 Nov 17 03:45 /usr/local/man/man1/xsetbg.1.gz -> /usr/local/man/man1/xli.1.gz
A commit references this bug: Author: gerald Date: Sun Dec 10 21:44:24 UTC 2017 New revision: 455914 URL: https://svnweb.freebsd.org/changeset/ports/455914 Log: Change man pages which are links (gitregrep.1.gz and gitrfgrep.1.gz) from absolute links to relative links. Bump PORTREVISION since this does affect the resulting package in a material manner. PR: 224230 Reported by: wosch Changes: head/misc/gnuit/Makefile head/misc/gnuit/files/patch-doc-Makefile
(In reply to Wolfram Schneider from comment #0) > I checked the packages on pkg.freebsd.org/freebsd:12:x86:64/latest/All > and found these 4 packages with absolute symlinks for man pages: That's actually pretty good for you to have only hit four packages! And if I did a good job, this should now be down to three? :-) (I have also submitted this upstream.) Thanks for the fine report, Wolfram!
A commit references this bug: Author: tobik Date: Thu Oct 11 14:33:19 UTC 2018 New revision: 481831 URL: https://svnweb.freebsd.org/changeset/ports/481831 Log: multimedia/libdca: Fix bad absolute symlinks PR: 224230 Reported by: wosch Changes: head/multimedia/libdca/Makefile head/multimedia/libdca/files/ head/multimedia/libdca/files/patch-src_Makefile.in
A commit references this bug: Author: tobik Date: Thu Oct 11 14:36:37 UTC 2018 New revision: 481832 URL: https://svnweb.freebsd.org/changeset/ports/481832 Log: graphics/xli: Fix bad absolute symlinks PR: 224230 Reported by: wosch Changes: head/graphics/xli/Makefile head/graphics/xli/files/patch-Imakefile
Should all be fixed now (ncbi-toolkit via ports r481472 from yuri@). But maybe there is something new now?
(In reply to Tobias Kortkamp from comment #5) I can re-run the test in a week to check if anything is fine.
(In reply to Wolfram Schneider from comment #6) Can I close?
I tested all packages again (72GB). The good news - all the known problems are fixed ;-)