If LOCALBASE is not /usr/local, the pango build fails because textproc/py-gi-docgen has a hard-coded reference to /usr/local.
/usr/loc/bin/gi-docgen generate --quiet --add-include-path=/wrkdirs/usr/ports/x11-toolkits/pango/work/pango-1.48.4/_build/docs/../pango --config docs/pango.toml --output-dir=docs/Pango --no-namespace-dir --content-dir=/wrkdirs/usr/ports/x11-toolkits/pango/work/pango-1.48.4/docs pango/Pango-1.0.gir
ERROR: Could not find GIR dependency in the search paths: GObject-2.0
Created attachment 225453 [details]
[patch] fix hard-coded /usr/local
Don't hard code the default search path with "/usr/local" - use LOCALBASE.
This patch bumps PORTREVISION. This patch will not change the default packages where LOCALBASE = /usr/local, so that bump is at the committer's discretion. I suggest it should be bumped - if someone has already installed a py-gi-docgen, but is failing the pango build because of a non-default LOCALBASE, it would help to bump PORTREVISION. In other words, this change does not directly fix a build error with the py-gi-docgen port, but rather a separate port that depends on py-gi-docgen. So bumping PORTREVISION is a good idea.
poudriere testport (textproc/py-gi-docgen, x11-toolkits/pango) - ok (stable/11 amd64)
portlint - ok
This has hit maintainer timeout. I wonder why the 'Problem reports' summary email for desktop@ does not include this bug.
A commit in branch main references this bug:
Author: John Hein <firstname.lastname@example.org>
AuthorDate: 2021-06-16 17:42:49 +0000
Commit: Tobias C. Berner <tcberner@FreeBSD.org>
CommitDate: 2021-06-16 17:42:59 +0000
textproc/py-gi-docgen: remove hard-coded /usr/local
textproc/py-gi-docgen/Makefile | 5 +++++
textproc/py-gi-docgen/files/patch-gidocgen_utils.py (new) | 11 +++++++++++
2 files changed, 16 insertions(+)