There're no well-known attributes

[im-kulikov]

Object and Container contains user-specific attributes ([]*Attribute is KeyValue) but there're no well-known attributes.

For example:

• CreatedAt - human-readable creation date/time (out of the box)
• Name - human-readable object/container name

@realloc, @alexvanin what are you think about adding constants for these attributes and the possibility to set them out of the box?

[realloc]

Let's add Name attribute to the well-known list in -api and -spec repos and then we can reflect that in api-go.

The CreatedAt attribute may be tricky. if it has Epoch, then IR can't set it because the whole structure is formed by owner and should not be changed. Owner can't predict for sure the Epoch of container being added into Container smart contract storage. We could get that information indirectly from the height of the block having container creation TX, but I'm not sure where to put it. Maybe in response meta headers, but that's strange.

If CreatedAt will have the unverified user defined date-time in some common format, then It's ok, but we need to name it somehow differently.

[im-kulikov]

In the closest future, I would like to see (SDK):

• Object.GetName() string
• Object.GetCreatedAt() time.Time
• Container.GetName() string
• Container.GetCreatedAt() time.Time

For example:
(Object/Container) -> Attributes -> Constant -> Result

---

PS: Using Epochs / BlockDate / TxDate seems like a bad idea for me

[alexvanin]

Don't mind to have those attributes. There is a good point that CreatedAt should be renamed to emphasize that this is unverified value. Maybe something like LocallyCreatedAt? Or LocalCreationTime.

[im-kulikov]

... CreatedAt should be renamed ... LocallyCreatedAt? Or LocalCreationTime

Naming is not as critical to me as the fact of having

[realloc]

Need to make sure neofs-api-go follows changes from nspcc-dev/neofs-api#77.

[im-kulikov]

Object:

• Filename
• Timestamp

Container:

• Name
• Timestamp

[realloc]

Object:

• Filename
• Timestamp

• Name

What did you mean?

[im-kulikov]

Object:

• Filename
• Timestamp

• Name

What did you mean?

For object we have two useful well-known attributes : filename and timestamp

This is notice for @alexvanin to add these attributes into sdk

Name isn’t useful attribute for object, that’s why I don’t add it into notice

For the container, we have next useful attributes: name and timestamp

[realloc]

Name isn’t useful attribute for object, that’s why I don’t add it into notice

Actually it's useful and has the same semantics and meaning as Name for Container. Filename should be used in cases like mirroring X-Filename http header to an object.

[im-kulikov]

Name isn’t useful attribute for object, that’s why I don’t add it into notice

Actually it's useful and has the same semantics and meaning as Name for Container. Filename should be used in cases like mirroring X-Filename http header to an object.

not quite.. for example, I can't find any useful usages for it in any of our services.

Filename, should be set by service and then, can be used to display for users, same as Timestamp

In this case, I don't see any useful usages for Name. As @alexvanin explained, this attribute can be used as human-readable replacement of the sha-hash.

And it can't be used the same way as Name for the container.