29f2157563
[ #123 ] protogen: Treat bytes field as non-nullable
...
In protobuf 3.12 they have added an support for `optional` keyword,
which has made it into the main branch in 3.15.
https://github.com/protocolbuffers/protobuf/blob/main/docs/implementing_proto3_presence.md
https://github.com/protocolbuffers/protobuf/blob/v3.12.0/docs/field_presence.md#presence-in-proto3-apis
This means that without an explicit `optional` keyword field presence
for scalars is not tracked, thus empty string in JSON should be
unmarshaled to a nil byte slice. Relevant decoding code and tests from
protojson:
fb995f184a/internal/impl/message_reflect_field.go (L327)
fb995f184a/encoding/protojson/decode_test.go (L134)
fb995f184a/encoding/protojson/decode_test.go (L156)
We do not support `optional` keyword and the generator will fail if it sees on.
So only implement the default behaviour.
Refs #122
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-10-11 14:40:54 +03:00
29c522d5d8
[ #122 ] protogen: Always marshal empty fields
...
This is how it was done previously:
a0a9b765f3/rpc/message/encoding.go (L31)
The tricky part is `[]byte` which is marshaled as `null` by easyjson
helper, but as `""` by protojson.
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-10-07 15:05:43 +03:00
eeb754c327
[ #120 ] protogen: Omit empty fields from JSON output
...
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-10-01 14:11:43 +03:00
5e1c6a908f
[ #111 ] protogen: Emit slice of messages without a pointer
...
```
goos: linux
goarch: amd64
pkg: git.frostfs.info/TrueCloudLab/frostfs-api-go/v2/refs
cpu: 11th Gen Intel(R) Core(TM) i5-1135G7 @ 2.40GHz
│ old │ new │
│ sec/op │ sec/op vs base │
ObjectIDSlice/0_elements/to_grpc_message-8 3.193n ± 2% 3.242n ± 0% +1.50% (p=0.034 n=10)
ObjectIDSlice/0_elements/from_grpc_message-8 3.197n ± 2% 3.343n ± 1% +4.57% (p=0.000 n=10)
ObjectIDSlice/0_elements/marshal-8 5.666n ± 3% 5.642n ± 0% -0.42% (p=0.000 n=10)
ObjectIDSlice/1_elements/to_grpc_message-8 53.10n ± 6% 29.78n ± 12% -43.92% (p=0.000 n=10)
ObjectIDSlice/1_elements/from_grpc_message-8 28.99n ± 5% 29.77n ± 7% ~ (p=0.165 n=10)
ObjectIDSlice/1_elements/marshal-8 49.08n ± 7% 50.72n ± 6% ~ (p=0.218 n=10)
ObjectIDSlice/50_elements/to_grpc_message-8 1652.5n ± 7% 277.2n ± 1% -83.22% (p=0.000 n=10)
ObjectIDSlice/50_elements/from_grpc_message-8 261.2n ± 11% 226.7n ± 15% -13.19% (p=0.003 n=10)
ObjectIDSlice/50_elements/marshal-8 1.512µ ± 6% 1.514µ ± 6% ~ (p=0.955 n=10)
geomean 52.15n 39.99n -23.31%
│ old │ new │
│ B/op │ B/op vs base │
ObjectIDSlice/0_elements/to_grpc_message-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/0_elements/from_grpc_message-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/0_elements/marshal-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/1_elements/to_grpc_message-8 32.00 ± 0% 24.00 ± 0% -25.00% (p=0.000 n=10)
ObjectIDSlice/1_elements/from_grpc_message-8 24.00 ± 0% 24.00 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/1_elements/marshal-8 48.00 ± 0% 48.00 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/50_elements/to_grpc_message-8 1.578Ki ± 0% 1.250Ki ± 0% -20.79% (p=0.000 n=10)
ObjectIDSlice/50_elements/from_grpc_message-8 1.250Ki ± 0% 1.250Ki ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/50_elements/marshal-8 2.000Ki ± 0% 2.000Ki ± 0% ~ (p=1.000 n=10) ¹
geomean ² -5.62% ²
¹ all samples are equal
² summaries must be >0 to compute geomean
│ old │ new │
│ allocs/op │ allocs/op vs base │
ObjectIDSlice/0_elements/to_grpc_message-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/0_elements/from_grpc_message-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/0_elements/marshal-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/1_elements/to_grpc_message-8 2.000 ± 0% 1.000 ± 0% -50.00% (p=0.000 n=10)
ObjectIDSlice/1_elements/from_grpc_message-8 1.000 ± 0% 1.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/1_elements/marshal-8 1.000 ± 0% 1.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/50_elements/to_grpc_message-8 51.000 ± 0% 1.000 ± 0% -98.04% (p=0.000 n=10)
ObjectIDSlice/50_elements/from_grpc_message-8 1.000 ± 0% 1.000 ± 0% ~ (p=1.000 n=10) ¹
ObjectIDSlice/50_elements/marshal-8 1.000 ± 0% 1.000 ± 0% ~ (p=1.000 n=10) ¹
geomean ² -40.18% ²
¹ all samples are equal
² summaries must be >0 to compute geomean
```
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-08-28 11:53:08 +03:00
5fece80b42
[ #108 ] protogen: Distinguish between empty and nil messages
...
Refs #59
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-08-27 11:02:18 +03:00
adb7c602d7
[ #77 ] protogen: Initial implementation
...
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-08-19 10:47:00 +03:00
3dfa2f4fd6
[ #95 ] *: Regenerate proto files
...
Remove SetExtendedEACL and AnnounceUsedSpace methods from the container
package.
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-07-26 10:24:25 +03:00
9789b79b1d
[ #85 ] apemanager: Generate protobufs for ape
package
...
* Regenerate protobufs as APE specific type are defined
as separate package;
* Move `ape` package related utils methods from `apemanager`.
Signed-off-by: Airat Arifullin <a.arifullin@yadro.com>
2024-05-29 14:15:49 +03:00
0fe6b9adbb
[ #76 ] Make setter names more consistent
...
This will make the transition to a new codegen more seamless.
1. Setter should be named as the field itself.
2. Setter should accept 1 parameter.
Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>
2024-05-03 08:52:04 +03:00
387b850e5e
[ #74 ] apemanager: Generate protobufs for apemanager service
...
* Generate protobufs.
* Make marshallers, unmarshallers, json-encoders etc.
* Create message encoding/decoding unit-tests.
Signed-off-by: Airat Arifullin <a.arifullin@yadro.com>
2024-05-02 11:53:26 +00:00