Re: groff-1.23.0.5077-7dcc8 on DragonFlyBSD 6.4.0, 6.4.1, and 6.4.2

"G. Branden Robinson" <[email protected]> Mon, 19 Jan 2026 18:44:47 -0600

Newsgroups	gmane.comp.printing.groff.general
Message-ID	<20260120004447.ejfcd6hqg7b3hjgx@illithid>

[CCing groff@gnu to keep developers in the loop]

Hi Nelson,

At 2026-01-19T17:28:21-0700, Nelson H. F. Beebe wrote:
> The build of groff-1.23.0.5077-7dcc8 on DragonFlyBSD 6.4.0, 6.4.1, and 6.4.2
> fails 5 tests:
> 
> XFAIL: tmac/tests/pdf_bookmark-starting-with-control-char-works.sh
> FAIL: src/preproc/tbl/tests/boxed-TH-tables-work-on-utf8-device.sh
> FAIL: src/roff/groff/tests/check-delimiter-validity.sh
> XFAIL: src/roff/groff/tests/current-language-and-environment-in-sync.sh
> XFAIL: src/roff/groff/tests/html-device-works-with-grn-and-eqn.sh
> XFAIL: src/roff/groff/tests/stringup-request-transforms-non-basic-latin.sh
> XFAIL: tmac/tests/e_footnotes-work-with-columns.sh
> # XFAIL: 5
> # FAIL:  2

You can ignore XFAILs.  They are "expected failures" under the test
harness Automake supports.  Once upon a time I knew the name of the RFC
or ISO standard that specified this testing scheme, but not anymore.

> The boxed-TH-tables-work-on-utf8-device.sh.log has numerous
> UTF-8 characters that are not readily handled by mail client.

Oh, darn.  Can you put this log file up at a paste-bin-style site?  I
personally use

https://paste.c-net.org/

> The check-delimiter-validity.sh.log is more than 600 lines long:
> here is a snippet:
> 
> troff:<standard input>:1: error: ignoring invalid numeric expression starting with character
> '_'
> ...
> troff:<standard input>:1: error: character '|' is not allowed as a delimiter
> troff:<standard input>:2: error: ignoring invalid numeric expression starting with nonexistent
> special character or class 'nl'
> troff:<standard input>:1: error: ignoring invalid numeric expression starting with character
> '@'

I'll need to see the whole log for that one, too.  This test is the most
savage in the whole suite (and has the longest runtime), and the
foregoing output doesn't tell me much because we deliberately test error
paths.  In other words, we _expect_ to see errors from troff.

The string "...FAILED" in full caps is how I mark the items of interest.
These should appear in any test script that "FAIL"s according to the
summary, and nowhere else.

I'll venture that the formatter is probably not really behaving
differently on DragonFlyBSD.  More likely, I've blundered into a
portability problem using echo(1) or printf(1) in my test script.

Please advise when you can get those 2 logs pastebinned.

Regards,
Branden

signature.asc (application/pgp-signature, 833 B)