Hi Hans, At 2026-09-29T16:16:47+0000, hbezemer--- via discussion of the GNU roff typesetting system and related software wrote: [...] > It seems that .INITR gives an error. [...] > cat > Test.groff << EOF > .INITR references > .TB "Table name" "" "" "Ref" > See Table > .GETST Ref > for more info. > EOF > mmroff -Tpdf Test.groff > Test.pdf
I've bisected the problem. $ git bisect good 7c8143d92af9132107a43f36f2916e06e6ad0aee is the first bad commit commit 7c8143d92af9132107a43f36f2916e06e6ad0aee Author: G. Branden Robinson <[email protected]> Date: Wed Aug 23 12:30:48 2023 -0500 [mm]: Validate reference system state and user input. * contrib/mm/m.tmac (qrf*validate-identifier): New macro tests argument for validity as roff identifier, since user input is applied to this purpose. (INITR): Throw error if mandatory argument omitted; keeps us from creating a (likely unwanted) dotfile, ".qrf". (SETR, GETST, GETHN, GETPN, GETR): Throw error if reference system macros used without initialization. (SETR, GETST, GETHN, GETPN): Validate arguments to be used as string identifiers. Also fix code style nits: Annotate logic. Stop uselessly calling `so` with no-break control character (it doesn't break in the first place). Break long input line. Use `\$0` in macro diagnostics instead of hard-coding caller. Fix misleading/unnecessary indentation. Simply control flow since `@error` does not return. contrib/mm/ChangeLog | 14 ++++++++ contrib/mm/m.tmac | 97 ++++++++++++++++++++++++++++++++++++++-------------- 2 files changed, 85 insertions(+), 26 deletions(-) Looks like not having sufficient unit tests for mmroff bit me. I'll file a Savannah ticket myself, but put you in the CC list. Regards, Branden
signature.asc
Description: PGP signature
