Changelog of all newer versions not in the tree: [1.4.2] 2023-02-17 Added ddcutil installation installs files /usr/lib/modules-load.d/ddcutil.conf and /usr/lib/modules-load.d#libddcutil.conf to ensure that kernel module i2c-dev is loaded at boot time if it is not built into the kernel. There are two files so that when split up into distribution packages, each of the command line ddcutil package and the shared library libddcutil package installs a file. [1.4.1] 2023-01-16 Fixed The default sleep-multipler value was 0, instead of 1. This resulted in failure of most DDC/CI operations, including display detection. [1.4.0] 2023-01-04 Added ddcutil installation installs file /usr/lib/udev/rules.c/60-ddcutil.rules. This udev rule uses tag uaccess to give the logged on user read/write access to /dev/i2c devices associated with video adapters. Configuring use of group i2c is no longer necessary. configure options --enable-syslog and --disable-syslog control whether messages are written to the system log. The default is enabled. Changed The ability to use the write()/read() interface of i2c-dev has been restored. It is needed to work around a bug in the proprietary Nvidia driver. By default, ioctl() interface is used for all drivers. If the Nvidia bug is detected, the write()/read() interface is used instead. Command line options --use-file-io and --use-ioctl-io affect this default behavior. When i2c-dev's file io interface is used, option --force-slave-address is again meaningful. Option --sleep-multiplier and API functions ddca_set_sleep_multiplier_value(), ddca_set_default_sleep_multiplier_value() now accept 0 as a valid argument. The ddcutil command parser reports an error if a display selection option (e.g. --bus) is given on a command to which it does not apply. Write additional error and information messages to the system log. Eliminate message "Is DDC/CI enabled in the monitor's on-screen display?" It's rarely the cause of communication failures. Fixed Warn of a possibly invalid DRM connector name in detect output if monitors with identical EDIDs are used with the proprietary nvidia driver. Handle /dev/i2c device names with a double hyphen, e.g. /dev/i2c--3. Better libddcutil handling of configuration file errors. Do not abort initialization in case of errors. Fix interpretation of digital display type bits for EDID version 1.4 Miscellaneous segfaults. [1.3.2] 2022-09-04 Changed Modify tarball creation to eliminate garbage and otherwise unneeded files. Reproducible: Always
Where do you see this 1.4.2 release? I see a 1.4.2 branch, but no tag or release.
(In reply to Robin Johnson from comment #1) > Where do you see this 1.4.2 release? I see a 1.4.2 branch, but no tag or > release. Hi direct on the release page of this util https://www.ddcutil.com/releases/ https://www.ddcutil.com/tarballs/ddcutil-1.4.2.tar.gz
There's no provenance to trace on those tarballs, and despite upstream's comment that they are identical, I see a lot more variation: source that is only in one set of tarballs and not the other. I'm going to stick w/ the tag-based tarballs on GitHub that can be proven as less possibly untampered. I did leave an upstream request asking for tags on the repo.
For 1.4.1, the file diff looks reasonable? It's just the difference between `make dist` and an autogenerated tarball of the git repo? What am I missing? Yes, the tags should be there, but the dist tarballs look fine and should be used in general.
In the diff: src/base/base.md | 8 ----- src/base/temp/tuned_sleep.h | 48 ++++++++++++++++++++++++++++++++++ src/util/systemd_util.c | 167 ------------------------------------------------------------------------------------------------------------------------- src/util/util.md | 11 ------- And the most likely place to sneak something bad in is m4 & configure scripts. 1.4.1 is good enough for now, 1.4.2 is for a niche behavior where i2c-dev should be loaded anyway already. Mostly waiting for the 2.0.0 to come out upstream.