[SPARK-50416][CORE] A more portable terminal / pipe test needed for bin/load-spark-env.sh #48937

philwalk · 2024-11-22T18:50:41Z

What changes were proposed in this pull request?

The last action in bin/load-spark-env.sh performs a test to determine whether running in a terminal or not, and whether stdin is reading from a pipe. A more portable test is needed.

Why are the changes needed?

The current approach relies on ps with options that vary significantly between different Unix-like systems. Specifically, it prints an error message in both cygwin and msys2 (and by extension, in all of the variations of git-for-windows). It doesn't print an error message, but fails to detect a terminal session in Linux and Osx/Darwin homebrew (always thinks STDIN is a pipe).

Here's what the problem looks like in a cygwin64 session (with set -x just ahead of the section of interest):

If called directly:

$ bin/load-spark-env.sh
++ ps -o stat= -p 1947
ps: unknown option -- o
Try `ps --help' for more information.
+ [[ ! '' =~ \+ ]]
+ [[ -p /dev/stdin ]]
+ export 'SPARK_BEELINE_OPTS= -Djline.terminal=jline.UnsupportedTerminal'
+ SPARK_BEELINE_OPTS=' -Djline.terminal=jline.UnsupportedTerminal'

Interestingly, due to the 2-part test, it does the right thing w.r.t. the Terminal test, the main problem being the error message.
If called downstream from a pipe:

$ echo "yo" | bin/load-spark-env.sh
++ ps -o stat= -p 1955
ps: unknown option -- o
Try `ps --help' for more information.
+ [[ ! '' =~ \+ ]]
+ [[ -p /dev/stdin ]]

Again, it correctly detects the pipe environment, but with an error message.

In WSL2 Ubuntu, the test doesn't correctly detect a non-pipe terminal session:

# /opt/spark$ bin/load-spark-env.sh
++ ps -o stat= -p 1423
+ [[ ! S+ =~ \+ ]]
# echo "yo!" | bin/load-spark-env.sh
++ ps -o stat= -p 1416
+ [[ ! S+ =~ \+ ]]

In #134-Ubuntu SMP Fri Sep 27 20:20:17 UTC 2024, the same failure occurs (it doesn't recognize terminal environments).

Does this PR introduce any user-facing change?

This is a proposed bug fix, and, other than fixing the bug, should be invisible to users.

How was this patch tested?

The patch was verified to behave as intended in terminal sessions, both interactive and piped, in the following 5 environments.


- Linux quadd 5.15.0-124-generic #134-Ubuntu SMP Fri Sep 27 20:20:17 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux
- Linux d5 5.15.153.1-microsoft-standard-WSL2 #1 SMP Fri Mar 29 23:14:13 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux
- MINGW64_NT-10.0-22631 d5 3.5.4-0bc1222b.x86_64 2024-09-04 18:28 UTC x86_64 Msys
- CYGWIN_NT-10.0-22631 d5 3.5.3-1.x86_64 2024-04-03 17:25 UTC x86_64 Cygwin
- Darwin suemac.local 23.6.0 Darwin Kernel Version 23.6.0: Mon Jul 29 21:14:21 PDT 2024; root:xnu-10063.141.2~1/RELEASE_ARM64_T8103 arm64

The test was to manually run the following script, verifying the expected response to both pipe and terminal sessions.

#!/bin/bash
if [ -e /usr/bin/tty -a "`tty`" != "not a tty" -a ! -p /dev/stdin ]; then
  echo "not a pipe"
else
  echo "is a pipe"
fi

The output of the manual test in all 5 tested environments.

philwalk@quadd:/opt/spark
$ isPipe
not a pipe
#
$ echo "yo" | isPipe
is a pipe
#

Was this patch authored or co-authored using generative AI tooling?

No

philwalk · 2024-11-22T19:02:18Z

Not sure what "workflow run detection" is, or how to correct it.
The output of git remote -vv is:

# git remote -vv
origin  https://github.com/philwalk/spark.git (fetch)
origin  https://github.com/philwalk/spark.git (push)

dongjoon-hyun

Thank you for making a PR, @philwalk .

Do you enable GitHub Action in your repository?

https://github.com/philwalk/spark/actions

There is a community documentation about Pull Request. Please check the section, Pull Request.

https://spark.apache.org/contributing.html

Go to “Actions” tab on your forked repository and enable “Build and test” and “Report test results” workflows

philwalk · 2024-11-24T21:16:18Z

I enabled Build and Test on my forked repo.

HyukjinKwon · 2024-11-25T00:01:40Z

Let's also file a JIRA, and add it into the PR description, see also https://spark.apache.org/contributing.html

pan3793 · 2024-11-25T05:42:58Z

The touched code seems to be borrowed from Apache Hive, would you like to also report the issue to the Hive community?

philwalk · 2024-11-25T17:31:55Z

Let's also file a JIRA, and add it into the PR description, see also https://spark.apache.org/contributing.html
I requested a JIRA account, not yet created ...

dongjoon-hyun · 2024-11-25T17:40:39Z

Got it. I approved it a few minutes ago, @philwalk .

philwalk · 2024-11-25T17:58:38Z

jira bug report

philwalk · 2024-11-25T17:59:19Z

The touched code seems to be borrowed from Apache Hive, would you like to also report the issue to the Hive community?
~~Not at this time ..~~
It seems that hive needs work to be compatible with cygwin or msys2 bash shell environments. There are a number of problems and I'm not able to work on it at this time.

By contrast, spark-shell starts right up (after applying this PR):

$ bin/spark-shell
WARNING: Using incubator modules: jdk.incubator.vector
Using Spark's default log4j profile: org/apache/spark/log4j2-pattern-layout-defaults.properties
Setting default log level to "WARN".
To adjust logging level use sc.setLogLevel(newLevel). For SparkR, use setLogLevel(newLevel).
Welcome to
      ____              __
     / __/__  ___ _____/ /__
    _\ \/ _ \/ _ `/ __/  '_/
   /___/ .__/\_,_/_/ /_/\_\   version 4.0.0-preview2
      /_/

Using Scala version 2.13.14 (OpenJDK 64-Bit Server VM, Java 21.0.4)
Type in expressions to have them evaluated.
Type :help for more information.
Spark context Web UI available at http://d5:4040
Spark context available as 'sc' (master = local[*], app id = local-1732575351925).
Spark session available as 'spark'.

scala>

philwalk · 2024-12-05T18:29:32Z

Is this waiting on me to do something?

github-actions · 2025-03-16T00:28:17Z

We're closing this PR because it hasn't been updated in a while. This isn't a judgement on the merit of the PR in any way. It's just a way of keeping the PR queue manageable.
If you'd like to revive this PR, please reopen it and ask a committer to remove the Stale tag!

bin/load-spark-env.sh

philwalk

LGTM

LuciferYang

+1, LGTM

LuciferYang · 2025-03-18T14:58:36Z

@dongjoon-hyun @HyukjinKwon @pan3793 Do you need to take another look?

philwalk · 2025-03-18T18:15:08Z

@dongjoon-hyun @HyukjinKwon @pan3793 Do you need to take another look?

I re-tested the modified line in MINGW64_NT-10.0-22631 d5 3.5.4-0bc1222b.x86_64 2024-09-04 18:28 UTC x86_64 Msys and the new version appears to be functionally identical, and the changes appear to be recommended style.

LuciferYang · 2025-03-18T18:21:24Z

I re-tested the modified line in MINGW64_NT-10.0-22631 d5 3.5.4-0bc1222b.x86_64 2024-09-04 18:28 UTC x86_64 Msys and the new version appears to be functionally identical, and the changes appear to be recommended style.

Thank you @philwalk , let's wait one more day. I will merge this one if there are no other new comments

…in/load-spark-env.sh ### What changes were proposed in this pull request? The last action in [bin/load-spark-env.sh](https://github.com/apache/spark/blob/d5da49d56d7dec5f8a96c5252384d865f7efd4d9/bin/load-spark-env.sh#L68) performs a test to determine whether running in a terminal or not, and whether `stdin` is reading from a pipe. A more portable test is needed. ### Why are the changes needed? The current approach relies on `ps` with options that vary significantly between different Unix-like systems. Specifically, it prints an error message in both `cygwin` and `msys2` (and by extension, in all of the variations of `git-for-windows`). It doesn't print an error message, but fails to detect a terminal session in `Linux` and `Osx/Darwin homebrew` (always thinks STDIN is a pipe). Here's what the problem looks like in a `cygwin64` session (with `set -x` just ahead of the section of interest): If called directly: ```bash $ bin/load-spark-env.sh ++ ps -o stat= -p 1947 ps: unknown option -- o Try `ps --help' for more information. + [[ ! '' =~ \+ ]] + [[ -p /dev/stdin ]] + export 'SPARK_BEELINE_OPTS= -Djline.terminal=jline.UnsupportedTerminal' + SPARK_BEELINE_OPTS=' -Djline.terminal=jline.UnsupportedTerminal' ``` Interestingly, due to the 2-part test, it does the right thing w.r.t. the Terminal test, the main problem being the error message. If called downstream from a pipe: ```bash $ echo "yo" | bin/load-spark-env.sh ++ ps -o stat= -p 1955 ps: unknown option -- o Try `ps --help' for more information. + [[ ! '' =~ \+ ]] + [[ -p /dev/stdin ]] ``` Again, it correctly detects the pipe environment, but with an error message. In WSL2 Ubuntu, the test doesn't correctly detect a non-pipe terminal session: ```bash # /opt/spark$ bin/load-spark-env.sh ++ ps -o stat= -p 1423 + [[ ! S+ =~ \+ ]] # echo "yo!" | bin/load-spark-env.sh ++ ps -o stat= -p 1416 + [[ ! S+ =~ \+ ]] ``` In `#134-Ubuntu SMP Fri Sep 27 20:20:17 UTC 2024`, the same failure occurs (it doesn't recognize terminal environments). ### Does this PR introduce _any_ user-facing change? This is a proposed bug fix, and, other than fixing the bug, should be invisible to users. ### How was this patch tested? The patch was verified to behave as intended in terminal sessions, both interactive and piped, in the following 5 environments. ``` - Linux quadd 5.15.0-124-generic #134-Ubuntu SMP Fri Sep 27 20:20:17 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux - Linux d5 5.15.153.1-microsoft-standard-WSL2 #1 SMP Fri Mar 29 23:14:13 UTC 2024 x86_64 x86_64 x86_64 GNU/Linux - MINGW64_NT-10.0-22631 d5 3.5.4-0bc1222b.x86_64 2024-09-04 18:28 UTC x86_64 Msys - CYGWIN_NT-10.0-22631 d5 3.5.3-1.x86_64 2024-04-03 17:25 UTC x86_64 Cygwin - Darwin suemac.local 23.6.0 Darwin Kernel Version 23.6.0: Mon Jul 29 21:14:21 PDT 2024; root:xnu-10063.141.2~1/RELEASE_ARM64_T8103 arm64 ``` The test was to manually run the following script, verifying the expected response to both pipe and terminal sessions. ```bash #!/bin/bash if [ -e /usr/bin/tty -a "`tty`" != "not a tty" -a ! -p /dev/stdin ]; then echo "not a pipe" else echo "is a pipe" fi ``` The output of the manual test in all 5 tested environments. ``` philwalkquadd:/opt/spark $ isPipe not a pipe # $ echo "yo" | isPipe is a pipe # ``` ### Was this patch authored or co-authored using generative AI tooling? No Closes #48937 from philwalk/portability-fix-for-load-spark-env.sh. Authored-by: philwalk <philwalk9@gmail.com> Signed-off-by: yangjie01 <yangjie01@baidu.com> (cherry picked from commit 8d26008) Signed-off-by: yangjie01 <yangjie01@baidu.com>

LuciferYang · 2025-03-20T02:35:59Z

Merged into master/branch-4.0/branch-3.5. Thanks @philwalk @dongjoon-hyun @HyukjinKwon @pan3793

more universal terminal / pip test for bin/load-spark-env.sh

175a03a

dongjoon-hyun reviewed Nov 22, 2024

View reviewed changes

HyukjinKwon changed the title ~~A more portable terminal / pipe test needed for bin/load-spark-env.sh~~ [SPARK-50416][CORE] A more portable terminal / pipe test needed for bin/load-spark-env.sh Nov 26, 2024

github-actions bot added the Stale label Mar 16, 2025

github-actions bot closed this Mar 17, 2025

LuciferYang reopened this Mar 17, 2025

LuciferYang removed the Stale label Mar 17, 2025

LuciferYang reviewed Mar 17, 2025

View reviewed changes

bin/load-spark-env.sh Show resolved Hide resolved

philwalk commented Mar 18, 2025

View reviewed changes

LuciferYang approved these changes Mar 18, 2025

View reviewed changes

LuciferYang closed this in 8d26008 Mar 20, 2025

[SPARK-50416][CORE] A more portable terminal / pipe test needed for bin/load-spark-env.sh #48937

[SPARK-50416][CORE] A more portable terminal / pipe test needed for bin/load-spark-env.sh #48937

Uh oh!

Conversation

philwalk commented Nov 22, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What changes were proposed in this pull request?

Why are the changes needed?

Does this PR introduce any user-facing change?

How was this patch tested?

Was this patch authored or co-authored using generative AI tooling?

Uh oh!

philwalk commented Nov 22, 2024

Uh oh!

dongjoon-hyun left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

philwalk commented Nov 24, 2024

Uh oh!

HyukjinKwon commented Nov 25, 2024

Uh oh!

pan3793 commented Nov 25, 2024

Uh oh!

philwalk commented Nov 25, 2024

Uh oh!

dongjoon-hyun commented Nov 25, 2024

Uh oh!

philwalk commented Nov 25, 2024

Uh oh!

philwalk commented Nov 25, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

philwalk commented Dec 5, 2024

Uh oh!

github-actions bot commented Mar 16, 2025

Uh oh!

Uh oh!

philwalk left a comment

Choose a reason for hiding this comment

Uh oh!

LuciferYang left a comment

Choose a reason for hiding this comment

Uh oh!

LuciferYang commented Mar 18, 2025

Uh oh!

philwalk commented Mar 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LuciferYang commented Mar 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

LuciferYang commented Mar 20, 2025

Uh oh!

Uh oh!

philwalk commented Nov 22, 2024 •

edited

Loading

dongjoon-hyun left a comment •

edited

Loading

philwalk commented Nov 25, 2024 •

edited

Loading

philwalk commented Mar 18, 2025 •

edited

Loading

LuciferYang commented Mar 18, 2025 •

edited

Loading