Telegraf: Autogenerated telegraf.conf uses UCS-2 LE-BOM coding

[antonkolb]

Relevant telegraf.conf:

autogenerated content

System info:

Windows 10 64-bit, x64 based processor

.\telegraf.exe --version shows
Telegraf 1.8.2 (git: HEAD 996a91f2)

Steps to reproduce:

1. Navigate to folder containing telegraf.exe
2. execute .\telegraf.exe config > telegraf.conf
3. execute .\telegraf.exe --config telegraf.conf

Expected behaviour:

No parsing error

Actual behaviour:

2018/10/18 15:32:31 E! Error parsing telegraf.conf, toml: line 1: parse error

Additional info:

Using PowerShell CLI

Opening file in Notepad++ and setting UTF-8 coding resolves the problem, as mentioned in issue #1378.

[antonkolb]

I tried this with MinGW64 (from Git Bash) and in this case the file is set to UTF-8 coding.

Sounds like a PowerShell issue.

Fun facts:
Size of the different files

text-compare.com shows a change in line 1, some changes regarding µ sign (and spaces in the lines before the inputs.jolokia.metrics) and an additional line at the end.

telegraf.zip

[danielnelson]

This might occur on other platforms as well if the terminal is set to a different locale, the proper behavior is to output in the encoding of stdout. I'm not sure if Go is handling this conversion or if Powershell did.

Separately it would be a nice touch to expand the file encodings we can read, not just with respect to the config file, but also reading text data formats.

[antonkolb]

Now I feel less silly reporting my issue. I assumed it is just PS bollocks and that I should have used something else in the first place.

[basroovers]

I implemented a workaround by catching the output in a variable and then out-file it with utf8 encoding.

$exe = "C:\Program Files\telegraf\telegraf.exe"
$telegrafConf = &$exe config
$telegrafConf | Out-File -FilePath "C:\Program Files\telegraf\telegraf.conf" -Encoding utf8