JSON object parsing "problem" in launch.json

[BorisPolonsky]

Issue Type: Bug

• Create the ill-foramted json object in launch.json as below:

{
    "version": "0.2.0",
    "configurations": [

        {
            "name": "Test training",
            "type": "python",
            "request": "launch",
            "program": "${cwd}/main.py",
            "console": "integratedTerminal",
            "env":{
                "LANG": "C.UTF-8"
            },
            "args": [
                "--mode", "train"
                "--data-dir", "./data/msg",
                "--output-dir", "./tmp_output",
                "--dataset", "some_dataset"
            ]
        }
    ]
}

(Note that a comma is missing after "train",, which means that the json object is invalid. I verify this with json.loads in python and got the following error: json.decoder.JSONDecodeError: Expecting ',' delimiter: line 16 column 17 (char 388))

• Execute the debugger

The debugging process runs without problem, instead of failing due to json parsing error. The args for debugger turn out to be correct too, given the terminal output like

root@10d189da27e6:/home/polonsky/Documents/EntityRelationExtraction# cd /home/polonsky/Documents/EntityRelationExtraction ; env LANG=C.UTF-8 PYTHONIOENCODING=UTF-8 PYTHONUNBUFFERED=1 /usr/bin/python3 /root/.vscode-server/extensions/ms-python.python-2020.2.64397/pythonFiles/ptvsd_launcher.py --default --client --host localhost --port 44049 /home/polonsky/Documents/EntityRelationExtraction/main.py --mode train --data-dir ./data/msg --output-dir ./tmp_output --dataset some_dataset

Please veryfy if this is a bug or some auto-correction feature.

Extension version: 2020.2.64397
VS Code version: Code 1.42.1 (c47d83b293181d9be64f27ff093689e8e7aed054, 2020-02-11T14:50:36.977Z)
OS version: Linux x64 5.3.0-40-generic snap
Remote OS version: Linux x64 5.3.0-40-generic

System Info

Item	Value
CPUs	Intel(R) Core(TM) i5-7300HQ CPU @ 2.50GHz (4 x 3100)
GPU Status	2d_canvas: unavailable_software flash_3d: disabled_software flash_stage3d: disabled_software flash_stage3d_baseline: disabled_software gpu_compositing: disabled_software metal: disabled_off multiple_raster_threads: enabled_on oop_rasterization: disabled_off protected_video_decode: disabled_off rasterization: disabled_software skia_renderer: disabled_off surface_control: disabled_off surface_synchronization: enabled_on video_decode: disabled_software viz_display_compositor: enabled_on viz_hit_test_surface_layer: disabled_off webgl: unavailable_software webgl2: unavailable_software
Load (avg)	2, 2, 1
Memory (System)	7.66GB (1.23GB free)
Process Argv	--force-user-env --no-sandbox --unity-launch --no-sandbox
Screen Reader	no
VM	0%

Item	Value
Remote	Container tensorflow/tensorflow:1.15.2-…
OS	Linux x64 5.3.0-40-generic
CPUs	Intel(R) Core(TM) i5-7300HQ CPU @ 2.50GHz (4 x 3100)
Memory (System)	7.66GB (1.23GB free)
VM	0%

[BorisPolonsky]

Please note that by adding : instead of ,, after "train", the debugger will run too.

[brettcannon]

We actually don't parse the launch.json, VS Code does on our behalf. It's also extended JSON which is much more lenient than standard JSON. Your best option would be to report this on the library that VS Code uses for parsing.