Developer Meeting Agenda 2026-11-08 - 10:00 CET

[grandixximo]

According to our bi-weekly schedule alternating between "early" and "late", this video meeting is Sunday 2026-Nov-08 at 10:00 CET.

https://greenlight.bbb.uni-rostock.de/b/ste-c4d-brs-3k6
Access code: 869782

https://www.timeanddate.com/worldclock/meetingdetails.html?year=2026&month=11&day=8&hour=9&min=0&sec=0&p1=319&p2=236&p3=240&p4=136&p5=165&p6=256

Agenda

0. Conclusions of 2026-10-25: nothing was posted in Developer Meeting Agenda 2026-10-25 - 20:00 CET #4705. Could someone who attended post them there, so the items below start from the right state?
1. Plan release 2.10, carried from 2026-10-11 item 13 and 2026-10-25. Decide: branch now, freeze master first, or branch once a named list is done, and which list? Since last time the HAL 64-bit component series from @BsAtHome landed (hal: Update hm2_modbus driver status params and pins to use 64-bit #4740, hal: Update streamer/sampler to use 64-bit accesses #4742 to hal: Update to 64-bit clean operation and accesses #4745, hal: Update debounce and modmath to run 64-bit clean #4772, hal: Update the pid component to 64-bit clean operation #4773; hal: Update sim_encoder to 64-bit clean operation #4771 sim_encoder still open).
2. NML replacement and the UIs built on it, discussion NML replacement #4571.
   • Out-of-tree UIs are appearing now. @rmu75 pointed at LiNC (Rust and egui, started 2026-10-22), which talks to LinuxCNC through linuxcnc-nml, Rust bindings over libnml covering the whole emcmodule.cc surface (all linuxcnc.command methods, all linuxcnc.stat fields, the error queue). On the forum there are also web UIs (lcncgui, lcnc-suite) on top of the Python module. Each of them breaks when NML goes.
   • Do we commit to keeping the client surface stable across the replacement: the Python linuxcnc module as it is, and maybe a C ABI client library with the same commands and status, so bindings like linuxcnc-nml can move over instead of being rewritten?
   • Requirements: @rene-dev wrote on 2026-10-28 that he has a good understanding of what is needed; @BsAtHome asked for it in detail. Who writes it down, and where (NML replacement #4571 or a new issue)?
   • Related open work: Zmq halui gui #4613 (ZMQ halui GUI), WIP: halui: pin-only GUI bridge surface (halui side of the #4613 split) #4616, Gui panel pins #4653 (GUI panel pins), modeless task, and jog-while-paused #4700 (modeless task, jog while paused), Unify the way jogging works from HAL, and its interaction with the UI #4692 (jogging from HAL), Unify the way MDI commands are executed from HAL / GUI #4691 (MDI from HAL and GUI).
3. Buildbot (buildbot2.highlab.com).
   • There are no Debian 13 trixie builders: the newest are bookworm (10-rip.debian-12-bookworm-rtpreempt-amd64, 50-deb-uspace.debian-12-bookworm-amd64, 30-dsc-uspace.debian-12-bookworm). GitHub CI builds trixie packages (package-arch, package-indep, 32-bit, rip-rtai), but runs the tests only on Ubuntu runners without a realtime kernel, and the buildbot is where the tests run under SCHED_FIFO. Can we get a trixie rtpreempt worker (rip with tests, deb, dsc)? Who has admin access to the buildbot and its hosts?
   • Retire the wheezy, jessie, stretch, buster and precise builders?
   • Two builders have failed in every one of their last 60 builds, back to at least 2026-07-22:
     • 50-deb-uspace.debian-12-bookworm-amd64: lintian errors executable-in-usr-lib on the qtvcp Python files (qtvcp/lib/qt_vismach/*.py, ripper/gcode_ripper.py, writer/writer.py).
     • 20-docs: the docs build, but the rsync to www.linuxcnc.org fails with Permission denied (publickey,password). Is the docs site still being updated, and whose key is it?
4. The wiki, discussion The wiki situation #4767 (carried from 2026-10-25): https is broken and nobody has admin access at dreamhost. Retire, move (dokuwiki was suggested), or fold what is still used into the docs? @andypugh named pages that only exist there (ContributedComponents, MinimumSoftwareVersions).
5. Forum (carried from 2026-10-25, mailing list thread "Critical PHP Vulnerabilities"): the forum is updated, still on Kunena. phpBB was floated on IRC; Discourse looks too heavy for the server. Stay on Kunena, or move, and who would do it?
6. Not reached on 2026-10-11, still open:
   • Development model, discussion Branch Protection #4558: areas and maintainers, what goes through a PR, cross-area changes, CI and branch protection, where it is written down.
   • Versioning and release procedure, CI build artifacts on master have wrong version #4555 and Rework versioning scripts and workflow #4584.
   • HAL names, Feature request: (Not)Accepting decoration as a hal signal name when there is no space #4645 (decoration at the end of a name) and hal - alphanumeric sort #4650 (natural sort in the tools).
   • mc_axis proposal, discussion Proposal: mc_axis — a PLCopen MC Part 1 axis component for HAL (extra joints, axes without G-code) #4647.
   • Logging library, discussion Logging library #4638.
   • Signed vs unsigned integers in HAL, The unsigned integer question #4526: who takes the conv_* components, the cross-net warning and the docs?
   • Meeting day and time (@rmu75 on the list).
7. HAL 64-bit, remaining:
   • Draft - HAL types update - Remove 32-bit and move to 64-bit #4247: close it, now that the break and the component series are in?
   • Converters hal: Add halfileupdate to convert HAL configuration files #4499 (halfileupdate) and hal: Add halcompupdate to migrate .comp files to the new HAL API #4256 (halcompupdate).
   • hal: C++ API (hal.hh) and pybind11 bindings on the new HAL API #4251 C++ API and pybind11 bindings, needed before branching 2.10 per 2026-10-11 item 4.
8. [DISPLAY]CYCLE_TIME, GUIs: read [DISPLAY]CYCLE_TIME with one rule #4677 (item 10 of 2026-10-11, decided on 2026-09-13). The PR implements the rule for every GUI and VCP, and following review writes seconds everywhere (73 sample configs move from 100 to 0.100). Behaviour changes to agree on:
   • The default becomes 0.150 s everywhere, so configs without CYCLE_TIME change: AXIS was 20 ms, QtVCP, GladeVCP and gscreen 100 ms, tklinuxcnc 200 ms.
   • Invalid values (zero, negative, below 0.001, non-numeric) now warn and fall back to the default.
9. PRs that change behaviour and need a decision:
   • interp, motion: the point-to-point moves and the tool orientation codes, G53.1 to G53.7 and G68.3 #4730 point-to-point moves and tool orientation codes, G53.1 to G53.7 and G68.3 (item 5 of the multiaxis roadmap Multiaxis kinematics and limits #4374). Large, no review yet: how do we review it?
   • motion: keep the last pose as the forward's guess after a failed forward #4757 motion: keep the last pose as the forward's guess after a failed forward (all KINEMATICS_BOTH machines).
   • homing: run every leg of a synchronized home sequence together #4577 homing: run every leg of a synchronized home sequence together.
   • xhc-hb04: rename dashed ini identifiers, migrate old configs in update_ini #4601 xhc-hb04: rename the dashed ini identifiers and migrate old configs in update_ini (follows the strict identifier decision). --This was decided earlier that the INI identifiers must be strict.--
   • preview: one preamble for every GUI, restating modes only #4618 preview: one preamble for every GUI, restating modes only. Needs a look from the GUI side.
   • 2.9: task: undo read-ahead modes when an error aborts a program (#4715) #4727 (undo read-ahead modes when an error aborts a program, RESET STATE IF NOT USE IN REMAP #4715), motion: drive joint.N.index-enable only on edges (backport #4200 to 2.9) #4204 (backport of motion: drive joint.N.index-enable only on edges (fix #3556) #4200, index-enable on edges), xhc-whb04b-6 fixes backport to 2.9 #4761 (xhc-whb04b-6 fixes backport, approved), then Merge branch '2.9' #4759 to merge 2.9 into master. Is another 2.9 release planned?
10. PRs waiting for a reviewer or an answer:
    • Canon cleanup #4748 canon cleanup.
    • xdr upstream readable #4762 XDR configure for macOS and other platforms.
    • hostmot2: fix PktUART queued counts and byte packing #4659, hostmot2: bound Ethernet packets and check transfers #4660, hostmot2: harden Smart Serial and absolute encoder handling #4662 hostmot2 PktUART, Ethernet and Smart Serial hardening. (still being reviewed)
    • motion: check circular moves against axis soft limits #4619 circular moves against soft limits (on hold, @rene-dev has a simpler idea).
    • tp: allow tangent blending across near-collinear ABC and UVW motion #4421 tangent blending across near-collinear ABC and UVW motion.
    • Add Support for Dogbone Homing #4009 dogbone homing.
    • Add generic and/or with flexibel inputs & allow unconnected pins #4746 and_n / or_n.
    • Realtime config scripts & doc #4752 realtime config scripts (draft).