Update all subprocess calls to use errors="backslashreplace"

[rmartin16]

Changes

• Update calls to subprocess to include errors="backslashreplace" to avoid UnicodeDecodeError exceptions
• Limit the ability for the output streamer to ignore ValueError exceptions to only those arising from the process unexpectedly closing stdout.

Example of effect of errors="backslashreplace"

Docker| [gtk/stdout] ‘girepository-1.0/Atk-1.0.typelib’ -> ‘/app/Hello World.AppDir//usr/lib/girepository-1.0/Atk-1.0.typelib’                          subprocess.py:641
Docker| [gtk/stdout] \xe2\x80[gtk/stderr] ++ realpath --relative-to=/usr/lib64 /usr/lib64/girepository-1.0                                              subprocess.py:641
Docker| [gtk/stderr] + cp girepository-1.0 --archive --parents '--target-directory=/app/Hello World.AppDir//usr/lib' --verbose                          subprocess.py:641
Docker| \x98girepository-1.0/SoupGNOME-2.4.typelib’ -> ‘/app/Hello World.AppDir//usr/lib/girepository-1.0/SoupGNOME-2.4.typelib’                        subprocess.py:641
Docker| [gtk/stdout] ‘girepository-1.0/GdkPixbuf-2.0.typelib’ -> ‘/app/Hello World.AppDir//usr/lib/girepository-1.0/GdkPixbuf-2.0.typelib’              subprocess.py:641

Issues

• Fixes Truncated Output or Killed Child Process When Output Streamer Suppresses UnicodeDecodeError Exceptions from popen_process.stdout.readline() #1141

PR Checklist:

• All new features have been tested
• All new features have been documented
• I have read the CONTRIBUTING.md file
• I will abide by the code of conduct

[rmartin16]

So, I thought through specific tests for this functionality....and I can't think of any that wouldn't simply be testing whether subprocess respects the errors setting or not (...presumably it does). In principle, this change should just affect whether subprocess raises UnicodeDecodeError for bytes it cannot decode....or prints them in hex (like in my example above).

If others have tests they think would be valuable, let me know and I'll add them.

[freakboy3742]

All looks good to me. One minor tweak to the formatting, mostly in service of making Black easier to deal with, but otherwise good to go.

I have a minor reservation about the sub_kw optimisation - it's a local test configuration we need to remember to use, and I don't necessarily consider verbosity a flaw in tests if the verbosity is explicit. However, I don't feel strongly enough to revert this myself (or ask you to revert it), and I can see some value in having a common definition so it's obvious and explicit when we're not using default values.

As for testing - I can't think of any obvious additional tests either. As you note, any testing of subprocess output really falls into the "testing that subprocess does what it says it does", which doesn't really improve quality.

[rmartin16]

Tbh, the consolidation was more about avoiding having to update 100 tests again in the future. I think this was the third time I've done it because something in Subprocess was changed.

[freakboy3742]

I figured as much :-) You will insist on working on gnarly bits of code, though... so you have no-one to blame by yourself :-P