https://bugs.documentfoundation.org/show_bug.cgi?id=173425
Pavel Řezníček <[email protected]> changed: What |Removed |Added ---------------------------------------------------------------------------- Ever confirmed|1 |0 Status|NEEDINFO |UNCONFIRMED --- Comment #5 from Pavel Řezníček <[email protected]> --- Tested against a copy of the actual broken profile from precision-7510 (the machine from the original report) -- never against the real one -- across several versions, per the request in comment 4. Method: official .deb tarball for each version, extracted to a non-system prefix (not installed), run with -env:UserInstallation pointing at a fresh copy of the profile that reliably crashes `unopkg add` with std::bad_alloc on 24.2.7.2 (same stale-registry state as in comment 3). Each test starts from that same broken state, removes the leftover unregistered entry, then retries `add`. Results: - 24.2.7.2 (originally reported) -- crashes, std::bad_alloc, as before - 24.8.7.2 -- succeeds (is registered: yes) - 25.2.7.2 -- succeeds - 25.8.7.3 -- succeeds - 26.2.6.3 -- succeeds - 26.8.1.1 -- succeeds So this is fixed by 24.8.7.2 at the latest, and stays fixed through the current 26.8.1.1. I did not bisect further inside the 24.2 branch, or try to pin down the exact fixing commit -- given 24.2 is already EOL, that didn't seem worth the effort, but happy to if it would help. (One side note in case it's confusing from the logs: the first 24.8.7.2 attempt failed too, but with a completely different, unrelated error -- our own extension's bundled Python dependency needing Python >=3.10, which 24.8's bundled Python (3.9) doesn't have. That's a bug in our own code, not LibreOffice's, and is now fixed on our side; mentioning it only so the version results above are understood as clean bad_alloc/no-bad_alloc outcomes.) -- You are receiving this mail because: You are the assignee for the bug.
