apache
Make opendal's path normalization logic more clear in public API
As reported from discord:
Shouldn't this be highlighted in the docs for bucket-type storages (correct me if already there)?
https://discord.com/channels/1081052318650339399/1201596970112917646
@Xuanwo As it stands, many otherwise valid read/write requests are deemed invalid due to path normalization. Do you think implementation of path normalization should be done differently for different services?
Do you think implementation of path normalization should be done differently for different services?
OpenDAL's vision is to access data freely, ensuring consistent user experience across various services.
Path normalization is useful in cases like writing a///b. On S3, it will be a///b, but on the file system, it will be a/b. It's very common for users to incorrectly join or concatenate paths.
OpenDAL does not aim to support use cases involving writing a///b to storage services. Thus I think we don't need to perform different path normalization on different services.