Commit Graph

8 Commits

Author SHA1 Message Date
Damien Neil
c5060d2fe6 reflect/protoreflect: add non-allocating Value constructors
Passing a non-pointer type to protoreflect.NewValue causes an
unnecessary allocation in order to store the value in an interface{}.
While this allocation could be avoided by a smarter compiler, no such
compiler exists today.

Add functions for creating new values of a specific type, avoiding the
allocation. (And also adding a small amount of type safety, although
this is unlikely to be important.)

Update the proto and internal/impl packages to use these functions.

Change-Id: Ic733de22ddf19c530189166c853348e1b54b7391
Reviewed-on: https://go-review.googlesource.com/c/protobuf/+/191457
Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>
2019-08-26 17:48:05 +00:00
Joe Tsai
0080e68288 reflect/protoreflect: improve package documentation
There are many types in this package, which can get very confusing how
they all relate to each other. Add some Unicode box art to show the
relationships between XType, XDescriptor, and X.

Change-Id: I935e645907270cdbb70c561e6cb394078366f861
Reviewed-on: https://go-review.googlesource.com/c/protobuf/+/190379
Reviewed-by: Damien Neil <dneil@google.com>
2019-08-20 18:44:49 +00:00
Joe Tsai
378c1329de reflect/protoreflect: add alternative message reflection API
Added API:
	Message.Len
	Message.Range
	Message.Has
	Message.Clear
	Message.Get
	Message.Set
	Message.Mutable
	Message.NewMessage
	Message.WhichOneof
	Message.GetUnknown
	Message.SetUnknown

Deprecated API (to be removed in subsequent CL):
	Message.KnownFields
	Message.UnknownFields

The primary difference with the new API is that the top-level
Message methods are keyed by FieldDescriptor rather than FieldNumber
with the following semantics:
* For known fields, the FieldDescriptor must exactly match the
field descriptor known by the message.
* For extension fields, the FieldDescriptor must implement ExtensionType,
where ContainingMessage.FullName matches the message name, and
the field number is within the message's extension range.
When setting an extension field, it automatically stores
the extension type information.
* Extension fields are always considered nullable,
implying that repeated extension fields are nullable.
That is, you can distinguish between a unpopulated list and an empty list.
* Message.Get always returns a valid Value even if unpopulated.
The behavior is already well-defined for scalars, but for unpopulated
composite types, it now returns an empty read-only version of it.

Change-Id: Ia120630b4db221aeaaf743d0f64160e1a61a0f61
Reviewed-on: https://go-review.googlesource.com/c/protobuf/+/175458
Reviewed-by: Damien Neil <dneil@google.com>
2019-06-17 17:33:24 +00:00
Damien Neil
a8593bae57 reflect/protoreflect: drop the ProtoEnum type
Drop the protoreflect.ProtoEnum type (containing a single method
returning a protoreflect.Enum) and make generated enum types
directly implement protoreflect.Enum instead.

Messages have a two-level type split (ProtoMessage and Message) to
minimize conflicts between reflection methods and field names. Enums
need no such split, since enums do not have fields and therefore have
no source of conflicts.

Change-Id: I2b6222e9404253e6bfef2217859e1b760ffcd29b
Reviewed-on: https://go-review.googlesource.com/c/156902
Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>
Reviewed-by: Damien Neil <dneil@google.com>
2019-01-09 00:40:35 +00:00
Joe Tsai
4b7aff630a all: rename Vector as List
The terminology Vector does not occur in protobuf documentation at all,
so we should rename the Go use of the term to something more recognizable.
As such, all instances that match the regexp "[Vv]ect(or)?" were replaced.

The C++ documentation uses the term "Repeated", which is a reasonable name.
However, the term became overloaded in 2014, when maps were added as a feature
and implementated under the hood as repeated fields. This is confusing as it
means "repeated" could either refer to repeated fields proper (i.e., explicitly
marked with the "repeated" label in the proto file) or map fields. In the case
of the C++ reflective API, this is not a problem since repeated fields proper
and map fields are interacted with through the same RepeatedField type.

In Go, we do not use a single type to handle both types of repeated fields:
1) We are coming up with the Go protobuf reflection API for the first time
and so do not need to piggy-back on the repeated fields API to remain backwards
compatible since no former usages of Go protobuf reflection exists.
2) Map fields are commonly represented in Go as the Go map type, which do not
preserve ordering information. As such it is fundamentally impossible to present
an unordered map as a consistently ordered list. Thus, Go needs two different
interfaces for lists and maps.

Given the above situation, "Repeated" is not a great term to use since it
refers to two different things (when we only want one of the meanings).
To distinguish between the two, we'll use the terms "List" and "Map" instead.
There is some precedence for the term "List" in the protobuf codebase
(e.g., "getRepeatedInt32List").

Change-Id: Iddcdb6b78e1e60c14fa4ca213c15f45e214b967b
Reviewed-on: https://go-review.googlesource.com/c/149657
Reviewed-by: Damien Neil <dneil@google.com>
2018-11-14 23:03:53 +00:00
Joe Tsai
812d9137df reflect/protoreflect: add Has and Clear to KnownFields and Map
Change the API to use explicit Has/Clear methods instead of relying on the
"invalid" form of Value to represent nullability.

There are several reasons for this change:

* Much of the ecosystem around protobufs do not care about nullability alone.
For example, for the encoder to determine whether to emit a field:
it needs to first check if a field is nulled, and if not, it still needs to go
through a series of type-assertion to check whether the value is the zero value.
It is much easier to be able to just call Has.

* It enables representing the default value as part of the value API, rather
than only as part of the descriptor API.

* The C++ API also uses explicit has and clear methods. However, we differ from
them by defining Has for proto3 scalars (while C++ panics instead).

For internal consistency, we also use a Has/Clear API for Maps.

Change-Id: I30eda482c959d3e1454d72d9fc761c761ace27a6
Reviewed-on: https://go-review.googlesource.com/134998
Reviewed-by: Damien Neil <dneil@google.com>
2018-09-13 19:52:11 +00:00
Damien Neil
2c8b670c6c reflect/protoreflect: put Value and MapKey types next to their methods
The Value and MayKey types are defined in value.go, but the methods on
these types are defined in value_union.go. It's not immediately clear
to the reader of the code that these types have methods, or where they
are defined.

Move the type definitions to value_union.go to keep the entire type in a
single place.

Change-Id: I7b3f3fc219a24a3b0236c2c3335e5d46f9086d25
Reviewed-on: https://go-review.googlesource.com/134997
Reviewed-by: Joe Tsai <thebrokentoaster@gmail.com>
2018-09-12 22:16:46 +00:00
Joe Tsai
dbc9a12dbd reflect/protoreflect: initial commit
Package protoreflect provides APIs for programatically interacting with messages
according to the protobuf type system.

Change-Id: I11fa3874e4b3f94771514c69efb2ae8bb812d7fa
Reviewed-on: https://go-review.googlesource.com/127823
Reviewed-by: Damien Neil <dneil@google.com>
2018-08-10 19:59:50 +00:00