iyad-f

@YoshiyukiMineo Nope, here are the logs ``` ? github.com/sony/sonyflake/v2/awsutil [no test files] ? github.com/sony/sonyflake/v2/example [no test files] ? github.com/sony/sonyflake/v2/mock [no test files] ? github.com/sony/sonyflake/v2/types [no test files] sonyflake id: 838915993...

i would like to mention a lot of those tests are dependent on how the hardware performs

there is no guarantee that only 1ms will pass from the start to the time when the id was generated, i tested multiple times, sometimes the diff is 1.63ms sometimes...

Ofcourse it will pass for larger timeunits even for 10ms it passes most of the times but again this isnt a perfect test, it comes with alot of dependencies, so...

This same thing can happen to TestNextID and TestNextID_InSequence. For TestNextID you have added a buffer of +1 to manage inconsistencies, but again depending upon the hardware this can change....

@kevinjqliu can i work on this issue ? I have already noted that we need an Enum and just replace format version everywhere with it?

@kevinjqliu I would have to replace it in classes of TableMetadata too right?, for example for TableMetadataV1 format_version would now be something like format_version: TableVersion = Field(alias="format-version", default=TableVersion.V1)

also wherever format_version is used as a value would have to do format_version.value, or would have to go the other way around for example for DATA_FILE_TYPE: Dict[int, StructType] it would...

naming the class TableVersion can be a bit consistent with what it was already, but at the same time error prone too if not noticed properly, since type checking is...

just a quick search for format_version in the library tells me it would need to be changed almost everywhere from tests to actual library code. If you have decided on...