Dvc: invalid start byte

[Sunny-Day200]

windows 10
dvc-0.92.0

git-bash 2.26.2.1

This is a very clean system
I cannot perform any DVC operations

Could it be because of the system environment?

---

ERROR: unexpected error - 'utf-8' codec can't decode byte 0x87 in position 0: invalid start byte

Having any troubles? Hit us up at https://dvc.org/support, we are always happy to help!
Traceback (most recent call last):
File "dvc__main__.py", line 7, in
File "dvc\main.py", line 90, in main
File "dvc\analytics.py", line 58, in is_enabled
File "dvc\config.py", line 235, in __init__
File "dvc\config.py", line 284, in load
File "dvc\config.py", line 297, in load_one
File "dvc\config.py", line 357, in _load_config
File "site-packages\configobj.py", line 1229, in __init__
File "site-packages\configobj.py", line 1287, in _load
File "site-packages\configobj.py", line 1485, in _handle_bom
File "site-packages\configobj.py", line 1517, in _decode
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x87 in position 0: invalid start byte
[30444] Failed to execute script __main__
[0x7FFD3CAFD3F0] ANOMALY: meaningless REX prefix used

---

And I want to ask about the cache link type, does it control the 'remote repository' and 'local cache' links?Or does it only work on 'local cache' links, and if it only works on 'local cache' links, can I change the 'DVC cache type, controlling the link type of the local cache data every time I pull git repository from a different platform?My English is not very good, I am not sure if I describe clearly, can you understand?

[efiop]

Hi @Sunny-Day200 !

Please show us $ dvc version output (it contains more than a version).

Correct, cache link type only controls local cache links. :slightly_smiling_face:

can I change the 'DVC cache type, controlling the link type of the local cache data every time I pull git repository from a different platform

Yes, you could use .dvc/config.local(used when you specify --local flag in your dvc config commands) for that. It is not tracked by git so it just stays in your local git repo instance.

[Sunny-Day200]

Thanked your reply

I tried it, but it didn't make any difference, right

$ dvc version
ERROR: unexpected error - 'utf-8' codec can't decode byte 0x87 in position 0: invalid start byte

Having any troubles? Hit us up at https://dvc.org/support, we are always happy to help!
Traceback (most recent call last):
File "dvc__main__.py", line 7, in
File "dvc\main.py", line 90, in main
File "dvc\analytics.py", line 58, in is_enabled
File "dvc\config.py", line 235, in __init__
File "dvc\config.py", line 284, in load
File "dvc\config.py", line 297, in load_one
File "dvc\config.py", line 357, in _load_config
File "site-packages\configobj.py", line 1229, in __init__
File "site-packages\configobj.py", line 1287, in _load
File "site-packages\configobj.py", line 1485, in _handle_bom
File "site-packages\configobj.py", line 1517, in _decode
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x87 in position 0: invalid start byte
[28480] Failed to execute script __main__
[0x7FFD3CAFD3F0] ANOMALY: meaningless REX prefix used

[karajan1001]

I think this is because your system uses a different encoding method (for example Chinese character may encode into gbk not utf-8 ). Do you use a pathname with some non-English character in it? If so, cleaning any non-English char in your pathname might solve your problem. But it's only a temporary solution, not a final one.

def _load_config(filename):
    conf_obj = configobj.ConfigObj(filename)
    return _parse_remotes(_lower_keys(conf_obj.dict()))

[karajan1001]

I can't reproduce your error. It seems that UTF8 encoding had been set to default since Windows 10 version 1903. Could you give me the result from the following python script?

import sys
print(sys.getfilesystemencoding())

[Sunny-Day200]

我无法重现您的错误。自Windows 10 1903版以来，似乎UTF8编码已设置为默认值。您能否从以下python脚本中获取结果？

import sys
print(sys.getfilesystemencoding())

It is indeed in utf-8 format. I haven't figured out what the problem is yet. It didn't happen on my other computers.I'm going to reinstall the system and try again

[karajan1001]

我无法重现您的错误。自Windows 10 1903版以来，似乎UTF8编码已设置为默认值。您能否从以下python脚本中获取结果？

import sys
print(sys.getfilesystemencoding())

It is indeed in utf-8 format. I haven't figured out what the problem is yet. It didn't happen on my other computers.I'm going to reinstall the system and try again

Have you tried using a folder name without any non-English characters?

[Sunny-Day200]

自Windows 10 1903版以来，似乎UTF8编码已设置为替换值。您能否从以下python脚本中获取结果？

import sys
print(sys.getfilesystemencoding())

它确实是utf-8格式。我还没有弄清楚问题出在哪里。在其他计算机上没有发生。我要重新安装系统，然后重试

您是否尝试使用没有任何非英文字符的文件夹名称？

The path does not contain any non-english characters

[Sunny-Day200]

I have some other questions

As described in the documentation, the hardlink, symlink link types make the workspace trace data file read-only (this prevents the user from accidentally corrupting the cache by modifying the file link).
As mentioned above, if I locally clone git from a remote location and change the copy of the cache link type to a symbolic link, is the trace file".dvc file" still not protected, and ".dvc file" will become read-only only after rebuilding "DVC add"
Sorry, I seem a little confused

[efiop]

@Sunny-Day200 That is expected behaviour. What is important is that the cache file in .dvc/cache stays read-only and is not corrupted. dvc add will create a new cache entry for a modified file and will link it as read-only back into your workspace.

[karajan1001]

I have some other questions

As described in the documentation, the hardlink, symlink link types make the workspace trace data file read-only (this prevents the user from accidentally corrupting the cache by modifying the file link).
As mentioned above, if I locally clone git from a remote location and change the copy of the cache link type to a symbolic link, is the trace file".dvc file" still not protected, and ".dvc file" will become read-only only after rebuilding "DVC add"
Sorry, I seem a little confused

Thank you, so strange. Maybe there is non-English char in your user name? And the error raises from loading global configuration.

[Sunny-Day200]

I have some other questions

As described in the documentation, the hardlink, symlink link types make the workspace trace data file read-only (this prevents the user from accidentally corrupting the cache by modifying the file link).
As mentioned above, if I locally clone git from a remote location and change the copy of the cache link type to a symbolic link, is the trace file".dvc file" still not protected, and ".dvc file" will become read-only only after rebuilding "DVC add"
Sorry, I seem a little confused

Thank you, so strange. Maybe there is non-English char in your user name? And the error raises from loading global configuration.

No , so I'm confused

[efiop]

Not able to reproduce :( Closing for now.

@Sunny-Day200 were you able to work around this issue?