El 30/9/26 a las 11:18, Ramiro Aceves escribió:
El mar, 29 sept 2026 a las 5:41, Ramiro Aceves (<ea1abz%gmail.com@localhost>) escribió:Hi Olaf, Thanks so much. At the moment is just a guess. Waiting for wednesday to see whether it fails and test the ctypeabuse! Regards. Ramiro.Today is wednesday (miércoles) and xscreensaver is working fine. My guess about ctype problem seems to be bad :-( . Waiting for the next failure (on saturday?, who knows). I have compiled and installed xscreensaver with debugging symbols just in case we can extract useful information. "é" is 0xc3a9= 11000011 10101001 "á" is 0xc3a1= 11000011 10100001 a9 and a1 are both "negative" and should crash ctype functions, is not it? Regards.
Hi,
Yes!, after observing xscreensaver during the whole the week, I can conclude that the malfunctioning
for unlocking the screen has only been observed on Saturdays in my system.
I found that the crash can be avoided in NetBSD by setting:
LIBC_ALLOWCTYPEABUSE=true xscreensaver
I also played with the locale settings. With:
LC_TIME="es_ES.UTF-8" -> fails
LC_TIME="C" -> works
LC_TIME="en_US.UTF-8" -> works
So far, this points to an interaction between xscreensaver, NetBSD ctype strict handling, and the Spanish LC_TIME locale.
For comparison, the same Saturday test on Linux and FreeBSD with Spanish locale did not show the problem.
Xscreensaver messages at the login unlock screen:
Linux:
2026-10-03 09:35:49
Spanish
FreeBSD:
2026-10-03 09:39:40
Spanish (Windows)
NetBSD with LIBC_ALLOWCTYPEABUSE="true":
09:58 a.m., sáb oct 3
Spanish
Regards,
Ramiro