Test output is truncated

[lilatomic]

Describe the bug
Output of tests is truncated.
Good:

15:47:01.49 [INFO] Completed: Run Pytest - src/python/pants/backend/helm/goals/lint_test.py:tests - succeeded.
============================= test session starts ==============================
--------8<-----------
src/python/pants/backend/helm/goals/lint_test.py::test_skip_lint PASSED  [100%]

- generated xml file: src.python.pants.backend.helm.goals.lint_test.py.tests.xml -
============================== 7 passed in 3.61s ===============================



✓ src/python/pants/backend/helm/goals/lint_test.py:tests succeeded in 4.13s.

Bad:

16:31:15.42 [INFO] Completed: Run Pytest - src/python/pants/backend/helm/goals/lint_test.py:tests - succeeded.
============================= test session starts ==============================
--------8<-----------
src/python/pants/backend/helm/goals/lint_test.py::test_skip_lint PASSED  [100%]


✓ src/python/pants/backend/helm/goals/lint_test.py:tests succeeded in 3.95s.

All the test output is captured, and I'm pretty sure it's being sent to be presented. I think it also has to do with the "vertical size" of the text, in that if I remove all the newlines from the message the full text is printed. There doesn't appear to be a major difference in behaviour between a shorter output and a much longer output, in that both have some truncation.

Pants version
2.19.0a0

OS
Linux

[lilatomic]

seems to have been something from 05e89ec
using --no-dynamic-ui, the output isn't truncated. I guess the output is being printed, but the dynamic UI covers it?
using --dynamic-ui-renderer=experimental-prodash also works fine.

[lilatomic]

reverting the upgrade of indicatif is sufficient to resolve this
specifically 0.17.4 is the breaking version

[tgolsson]

Can you isolate a repro? I haven't observed anything like this and I did a bunch of testing.

[tgolsson]

Also, can you see it happening with anything except test output?

[kaos]

this issue was also identified by @stuhood in #19842

[tgolsson]

Yeah. I've looked a bit at the indicatif codebase and if we can find a repro I can probably fix it. But I've been running with that fix locally and in multiple projects without seeing any issues. It does seem like a race-like problem so maybe dependent on load etc.

[lilatomic]

I tried with the other goals. I think it affects them too, but it's harder to tell (lint and check are missing a few blank lines at the bottom).
The repro steps are just "do anything", so I think it might be environment dependent. I was trying in a sideproject, in the pants repo I was trying ./pants test --force --output=all src/python/pants/backend/helm/goals/lint_test.py to generate the above. It's consistent across shells (bash, fish) and terminals (gnome terminal, intellij's builtin, vscode's builtin).
System info:

uname -a
Linux MS-7A34 6.2.0-36-generic #37~22.04.1-Ubuntu SMP PREEMPT_DYNAMIC Mon Oct  9 15:34:04 UTC 2 x86_64 x86_64 x86_64 GNU/Linux

lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 22.04.3 LTS
Release:        22.04
Codename:       jammy

[tgolsson]

Do you have any other setting overrides apart from that command line?

My output is consistently the following (>10 tries):

./pants test --force --output=all src/python/pants/backend/helm/goals/lint_test.py
11:54:47.35 [INFO] Completed: Run Pytest - src/python/pants/backend/helm/goals/lint_test.py:tests - succeeded.
============================= test session starts ==============================
collecting ... collected 7 items

src/python/pants/backend/helm/goals/lint_test.py::test_lint_non_strict_chart_passing PASSED [ 14%]
src/python/pants/backend/helm/goals/lint_test.py::test_lint_non_strict_chart_failing PASSED [ 28%]
src/python/pants/backend/helm/goals/lint_test.py::test_lint_strict_chart_failing PASSED [ 42%]
src/python/pants/backend/helm/goals/lint_test.py::test_global_lint_strict_chart_failing PASSED [ 57%]
src/python/pants/backend/helm/goals/lint_test.py::test_lint_strict_chart_passing PASSED [ 71%]
src/python/pants/backend/helm/goals/lint_test.py::test_one_lint_result_per_chart PASSED [ 85%]
src/python/pants/backend/helm/goals/lint_test.py::test_skip_lint PASSED  [100%]


✓ src/python/pants/backend/helm/goals/lint_test.py:tests succeeded in 2.18s.

So that has multiple differences to yours, and seems (a) correct and (b) consistent.

[lilatomic]

that output should have the lines

- generated xml file: src.python.pants.backend.helm.goals.lint_test.py.tests.xml -
============================== 7 passed in 3.61s ===============================

I don't have any other overrides in place.

[tgolsson]

Ack, thanks. Got confused specifically by this part in the original post: --------8<----------- and having fewer tests. Just wanted to be sure we're comparing apples-to-apples. So I'm reproducing the wrong output every time. That's a good start. I'll have a bit more of a dig tomorrow but if I can't find a solution I'll revert the upgrade.

It's interesting, the reason I upgraded was for some other multiline output to work correctly -- streaming workunit logs into each job. That is broken on 0.17.4... 😢

Edit: And I have a repro outside of Pants! The issue is \n\n! That trips up indicatif space counting somehow.
Edit: Ok, as expected that was fairly obvious once a repo was isolated, indicatif is a nice codebase. Will see if upstream responds, otherwise we'll have to do like with console and use a fork.