Bug 215077 – run-jsc-stress-test: Set DYLD_FRAMEWORK_PATH to "$(cd ../.vm; pwd)" because it escape shell words

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Zhifei Fang]

run-jsc-stress-test: Set DYLD_FRAMEWORK_PATH to "$(cd ../.vm; pwd)" because it escape shell words

[Zhifei Fang]

Created attachment 405819 [details]
Patch

[Zhifei Fang]

rdar://66451802

[Zhifei Fang]

break by this change:

https://bugs.webkit.org/show_bug.cgi?id=214356

[Zhifei Fang]

Hi, Angelos

Any special reason to escape the DYLD_FRAMEWORK_PATH ?

[EWS]

Committed r265204: <https://trac.webkit.org/changeset/265204>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 405819 [details].

[Radar WebKit Bug Importer]

<rdar://problem/66489295>

[Caio Lima]

(In reply to Zhifei Fang from comment #4)
> Hi, Angelos
> 
> Any special reason to escape the DYLD_FRAMEWORK_PATH ?

Hi Zhifei. Angelos is on holidays now and he couldn't answer it on time. This change caused some issues to gnu-parallel-runner configuration (It is possible to see it on EWS ARMv7 and MIPS run):

```
/bin/bash: -c: line 0: syntax error near unexpected token `('
/bin/bash: -c: line 0: `cd jsc-stress-results/.runner && export DYLD_FRAMEWORK_PATH=\$(cd ../.vm; pwd) && export LD_LIBRARY_PATH=\$(pwd)/jsc-stress-results/.vm/JavaScriptCore.framework/Helpers &&export JSCTEST_timeout=1200 && export JSCTEST_hardTimeout=300 && export JSCTEST_memoryLimit=838860800 && export TZ=US/Pacific && sh  ./test_script_25617'
```

IIUC, this change was necessary to allow Make runner work again for remote devices, right? I'm not very familiar with this part of the code and with pearl, so I don't want to make too much assumptions there.

In the meantime, I changed configuration of those bots to use Make runner instead of gnu-parallel, so we can process EWS and post-commit bots queues.

[Angelos Oikonomopoulos]

(In reply to Zhifei Fang from comment #4)
> Hi, Angelos
> 
> Any special reason to escape the DYLD_FRAMEWORK_PATH ?

Hi Zhifei,

the reason was I was using different kind of quoting when invoking GNU parallel. See https://bugs.webkit.org/show_bug.cgi?id=215887 for a suggested fix.