[tankf33der]

Latest supported Alpine Linux version is 3.18.12, it have gettid() in manual pages.
Now test is broken, lets make it work. Tested on latest 3.21.3

94cc75377785:~/v/vlib# ../v test v/tests/trace_calls_test.v
---- Testing... ----------------------------------------------------------------------------------------------------------------------------
 FAIL  5033.210 ms /root/v/vlib/v/tests/trace_calls_test.v
> trace_fns_only_println compilation output:
================== C compilation error (from cc): ==============
cc: /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8FW0CN09699MYB2GPFW8.tmp.c:2120:31: error: conflicting types for 'gettid'; have 'u32 *(void)' {aka 'unsigned int *(void)'}
cc:  2120 | VV_EXPORTED_SYMBOL VWEAK u32* gettid(void); // exported fn v.trace_calls.vgettid
cc:       |                               ^~~~~~
cc: In file included from /usr/include/fortify/unistd.h:23,
cc:                  from /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8FW0CN09699MYB2GPFW8.tmp.c:507:
cc: /usr/include/unistd.h:198:7: note: previous declaration of 'gettid' with type 'pid_t(void)' {aka 'int(void)'}
cc:   198 | pid_t gettid(void);
cc:       |       ^~~~~~
cc: /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8FW0CN09699MYB2GPFW8.tmp.c: In function 'v__trace_calls__on_call':
cc: /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8FW0CN09699MYB2GPFW8.tmp.c:14527:29: warning: assignment to 'u32' {aka 'unsigned int'} from 'u32 *' {aka 'unsigned int *'} makes integer from pointer without a cast [-Wint-conversion]
cc: 14527 |                         tid = gettid();
cc:       |                             ^
...
cc:       |       ^~~~~~
(note: the original output was 19 lines long; it was truncated to its first 12 lines + the last line)
================================================================
(You can pass `-cg`, or `-show-c-output` as well, to print all the C error messages).
builder error:
==================
C error found. It should never happen, when compiling pure V code.
This is a V compiler bug, please report it using `v bug file.v`,
or goto https://github.com/vlang/v/issues/new/choose .
You can also use #help on Discord: https://discord.gg/vlang .

/root/v/vlib/v/tests/trace_calls_test.v:55: fn run
   > assert res.exit_code == 0, 'compilation of $fpath failed'
     Left value (len: 1): `1`
    Right value (len: 1): `0`
        Message: compilation of vlib/v/tests/testdata/trace_calls/single_println.vv failed

> 64bit compilation output:
================== C compilation error (from cc): ==============
cc: /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8JEMC71Y50G8HM1EHM6M.tmp.c:2120:31: error: conflicting types for 'gettid'; have 'u32 *(void)' {aka 'unsigned int *(void)'}
cc:  2120 | VV_EXPORTED_SYMBOL VWEAK u32* gettid(void); // exported fn v.trace_calls.vgettid
cc:       |                               ^~~~~~
cc: In file included from /usr/include/fortify/unistd.h:23,
cc:                  from /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8JEMC71Y50G8HM1EHM6M.tmp.c:507:
cc: /usr/include/unistd.h:198:7: note: previous declaration of 'gettid' with type 'pid_t(void)' {aka 'int(void)'}
cc:   198 | pid_t gettid(void);
cc:       |       ^~~~~~
cc: /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8JEMC71Y50G8HM1EHM6M.tmp.c: In function 'v__trace_calls__on_call':
cc: /tmp/v_0/tsession_4002904cc0_01JRYX8ABNH5HRE009SG5GQZVM/single_println.01JRYX8JEMC71Y50G8HM1EHM6M.tmp.c:15212:29: warning: assignment to 'u32' {aka 'unsigned int'} from 'u32 *' {aka 'unsigned int *'} makes integer from pointer without a cast [-Wint-conversion]
cc: 15212 |                         tid = gettid();
cc:       |                             ^
...
cc:       |       ^~~~~~
(note: the original output was 19 lines long; it was truncated to its first 12 lines + the last line)
================================================================
(You can pass `-cg`, or `-show-c-output` as well, to print all the C error messages).
builder error:
==================
C error found. It should never happen, when compiling pure V code.
This is a V compiler bug, please report it using `v bug file.v`,
or goto https://github.com/vlang/v/issues/new/choose .
You can also use #help on Discord: https://discord.gg/vlang .

/root/v/vlib/v/tests/trace_calls_test.v:55: fn run
   > assert res.exit_code == 0, 'compilation of $fpath failed'
     Left value (len: 1): `1`
    Right value (len: 1): `0`
        Message: compilation of vlib/v/tests/testdata/trace_calls/single_println.vv failed
--------------------------------------------------------------------------------------------------------------------------------------------
To reproduce just failure 1 run:    '/root/v/v'  '/root/v/vlib/v/tests/trace_calls_test.v'
Summary for all V _test.v files: 1 failed, 1 total. Elapsed time: 10533 ms, on 1 job. Comptime: 5474 ms. Runtime: 5031 ms.

[huly-for-github]

Connected to Huly®: V_0.6-22627

[tankf33der]

@spytheman - how to make it work on musl:Alpine and skip musl:Ubuntu?
is it valid place for this?

[spytheman]

@spytheman - how to make it work on musl:Alpine and skip musl:Ubuntu? is it valid place for this?

Currently there is no way for detecting that.

[spytheman]

If only vlib/v/tests/trace_calls_test.v is failing, we can add a build tag at its start, that says:

// vtest build: !docker-ubuntu-musl

... which will skip it for only that CI job, but it will still fail for people that are running that test outside the CI.

[tankf33der]

Then my goal to make vlib/v/tests without errors on s390x on Alpine in my own CI is unachievable.
If there is no other way, then cancel this PR and I'll think of something else

[spytheman]

Then my goal to make vlib/v/tests without errors on s390x on Alpine in my own CI is unachievable. If there is no other way, then cancel this PR and I'll think of something else

Why?
My suggestion is to edit the vlib/v/tests/trace_calls_test.v file, and add the comment that I showed.

[spytheman]

Your own CI can also set an env variable with the name of the job.
See cmd/v/v.v:228 for the handling of GITHUB_JOB, that allows for // vtest build: !docker-ubuntu-musl to work as a filter later.

[tankf33der]

Then my goal to make vlib/v/tests without errors on s390x on Alpine in my own CI is unachievable. If there is no other way, then cancel this PR and I'll think of something else

Why? My suggestion is to edit the vlib/v/tests/trace_calls_test.v file, and add the comment that I showed.

Lost in translation.
I will do it today.

[spytheman]

You can also test for musl in those tags/comments and all other $if platform_identifier {, just not for the specific version of musl directly at compile time.

[spytheman]

I suppose that we can add detection for the distro too, since that can be of general use for other things as well.

[spytheman]

I've rebased it over the current master (which now has a way to detect the distro too for v test runs). Lets see if the CI will pass this time.

[spytheman]

It passed: