All types of attribute keys must be unique and can't be repeated in the same
entity.
- Containers with duplicated attribute keys must not be accepted by InnerRing on
creation.
- Nodes with duplicated attribute keys can't be accepted to NetMap by InnerRing
- Objects with duplicated attribute keys must be considered invalid and not
accepted in PUT operations
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
SplitInfo structure from split_info field contains meta information
to assembly the object that has been split into parts.
Signed-off-by: Alex Vanin <alexey@nspcc.ru>
In case there is a need to set flag-like attributes, the value should be set to
something like `true` or `1`.
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
In some cases like nspcc-dev/neofs-node#167 there is no simple way to identify
where does the object part belongs to. Adding `split_id` as the required field
will simplify split object processing.
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
NeoFS uses different checksum algorithms. It looks like we need to explicitly
define what algorithm is used in each particular case.
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
Some pieces of data (container, object, etc) may be stored for a long time and
there will be a need in the future to understand which obscure format from the
past was used to create it.
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
To simplify adding more transport level protocols to neofs-api-go in future, we
need to separate currently default gRPC.
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
- Changed package names adding version
- Added documentation descriptions (sometimes useless) for all fields
- Changed enum format
- Made SessionToken and BearerToken field names more clear
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
NeoFS Object are now Content-addressed. It means the Object's address depends on
it's content. ObjectID is now calculated as hash of Header, which contains a
hash of payload. If either if Object's payload of Headers change, the ID will
also change.
Signed-off-by: Stanislav Bogatyrev <stanislav@nspcc.ru>
Change the type of all fields for identifier of the container owner to
refs.OwnerID. This will allow you to follow a single format and not duplicate
its description.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
In previous version of the format storage group information was stored in
completely in the dedicated header field. However, having a header field
instead of storing it in a payload is not justified, since it is used only
to check the presence in the object. Based on this, this commit defines a
message for marking the object as the custodian of information about the
storage group and changes the type of field StorageGroup to it. Information
about the storage group will be stored serialized in the payload of the
objects marked in this way
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
In the previous format, the object header was subdivided into system and
extended parts, which confused the purpose of these parts. In particular,
the extended header contains system fields. To clarify the non-intersection
of the mentioned parts of the object header, it was decided to rename System
to Main.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
This commit replaces Token field from ExtendedHeader to IntegrityHeader for
assembling all verification data in one structure.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
This commit replaces single Attribute field with the repeated list of
Attribute in ExtendedHeader message.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
This commit replaces PayloadChecksum field from ExtendedHeader to
IntegrityHeader for assembling all verification data in one structure.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
This commit moves the public key of the creator of the object to
IntegrityHeader to encapsulate the verification data in one message. Thus
field PublicKey of message Header has been moved to message IntegrityHeader
with the name CreatorKey. As a result, PublicKey message is deleted as no
longer used.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
Redirect field was introduced for the future, however, at the moment it is
not used anywhere. This commit removes field Redirect from ExtendedHeader
message. It can be added back later without losing API backward
compatibility.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
After updating the structure of SplitHeader and StorageGroup messages, the
Link header field is no longer needed. This commit removes the field and
message Link as no longer used.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
Field with object revision number was introduced for the future, however, at
the moment it is not used anywhere. This commit removes field Version from
SystemHeader message. It can be added back later without losing API
backward compatibility.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
In previous version of the format ExtendedHeader consisted of one field
Value, defined by oneof mechanism. Such a definition format give complete
freedom when composing an object header, however, it did not allow strictly
structuring the header format at the level of type syntax. Since the object
has a well-defined structure in the system, this commit replaces the
definition through oneof with a set of fields. As a consequence, storing a
set of extended headers of the new type in the general header became
redundant, so the repeated keyword was removed.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
This commit replaces Transform message with SplitHeader in oneof definition
of Value field of Header message. The latter, unlike the first, carries
complete information about the generation of the object by splitting the
payload. As a result, the message is deleted as no longer used.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
This commit defines SplitHeader message that encapsulates data about
spawning the object through a payload splitting. Message fields contain
information about the hierarchy in the split-chain as well as information
about the origin object.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
This commit merges object system header fields ID and CID into field Address
of type refs.Address. This will allow you to reuse an already existing type
of object reference without duplication and simplify taking the address
from the object body
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
In previous version of the format Object message included moment of creation
in two epoch times: NeoFS and Unix. Since unix is not processed in the
system, this commit narrows the time point of object creation to the system
epoch number. The object.CreationPoint message has therefore been deleted
as no longer used
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
In previous version of the format Object message included two header fields:
SystemHeader and list of ExtendedHeader. Due to the fact that all headers
were not encapsulated in a single message, some messages were forced to use
Object to convey the full header of the object (e.g. object.HeadResponse).
To resolve this semantic inconsistency, this commit defines a Header
message, consisting of a system and an extended portion. As a result, the
Object now consists of two fields - Header and Payload.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
