On Sun, Dec 31, 2023 at 10:33:36AM -0800, Edward Chron wrote:
> + if (*rec->caller_id) {
> + if (ctl->json) {
> + ul_jsonwrt_value_s(&ctl->jfmt, "caller", rec->caller_id);
> + } else {
> + char cidbuf[PID_CHARS_MAX+3] = {'\0'};
> +
> + sprintf(cidbuf, "[%*s] ",
> + (char)ctl->caller_id_size, rec->caller_id);
> + ctl->indent += strnlen(cidbuf, sizeof(cidbuf));
> + fputs(cidbuf, stdout);
> + }
> + }
The variable width (ctl->caller_id_size) of caller ID makes your
regression test fragile, see:
https://github.com/util-linux/util-linux/actions/runs/7384780996/job/20088287790?pr=2647
--- /home/runner/work/util-linux/util-linux/tests/expected/dmesg/cid-limit 2024-01-02 10:34:02.893193174 +0000
+++ /home/runner/work/util-linux/util-linux/tests/output/dmesg/cid-limit 2024-01-02 10:38:29.209015303 +0000
@@ -1,4 +1,4 @@
-[ 1.000000] [ T1] example[1]
-[ 8.000000] [ T2] example[2]
-[ 27.000000] [ T3] example[3]
-[ 64.000000] [ T4] example[4]
+[ 1.000000] [ T1] example[1]
+[ 8.000000] [ T2] example[2]
+[ 27.000000] [ T3] example[3]
+[ 64.000000] [ T4] example[4]
I see two possible ways to fix it:
* "normalize" the output in the tests -- just use sed(1) to
remove all the blanks space "[ T4]" to "[T4]", so the output
will be always the same
* or use fixed width for the caller_id in dmesg.c
Karel
--
Karel Zak <kzak@xxxxxxxxxx>
http://karelzak.blogspot.com
