Le samedi, 6 décembre 2025, 13.35:27 h heure normale d’Europe centrale Sebastian Humenda a écrit : > Didier 'OdyX' Raboud schrieb am 05.12.2025, 9:52 +0100: > >I have fixed these two bugs in the incoming NMU 1.13.1+dfsg-1.2 on dictd, > >via allowing burst restarts of the dictd service. > > What was the necessity to NMU these? Could you open a PR on Salsa?
Thank you for asking Sebastian, The original 1.13.1+dfsk-1.1 NMU happened to fix #1039171 which was serious since October 15. I uploaded it to DELAYED/2 after a series of patches were proposed to me in private by Kevin. Then I filed a merge-request here: https://salsa.debian.org/debian/dictd/-/merge_requests/4 After it landed in unstable, unforeseen consequences started to appear, namely via #1066189, #1121708 and via failing autopackagetests (blocking the migration to forky). Together with Kevin (original proposer of the -1.1 NMU), we followed-up with the -1.2 NMU fixing some of the systemd integration. But as Bertram mentionned in #1121708, I was not cautious enough when adding the systemd-unit StartLimitIntervalSec=0 line, so this needs a new -1.3 NMU to fix this. Subsequent MRs (one per NMU) are: https://salsa.debian.org/debian/dictd/-/merge_requests/5 https://salsa.debian.org/debian/dictd/-/merge_requests/6 It would be really nice to get input from the package maintainer (CC'ed in this email, and should receive the bts traffic), but from what I can gather (notably from https://contributors.debian.org/contributor/robert/), they were not active on packages in 2025 at all. I'll continue to NMU dictd as much as needed to fix the regressions introduced by the systemd unit migration. But, as mentioned earlier: the fact that providers of dictionaries all restart dictd (whether in initd or systemd) is antique and should be implemented via dpkg triggers (#1121960). The migration to systemd just made the issue of dictd restarting in bursts during upgrades just way more visible. Best, -- OdyX
signature.asc
Description: This is a digitally signed message part.
