river/internal/value: change representation of objects and maps #2369
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
rfratto
force-pushed
the
river-capsule-type-loss-fix
branch
from
720d358 to
199b9f8
Compare
Add new benchmarks to test a fix for grafana#2362 and ensure that the benchmarks for stdlib are using a sufficiently complicated structure that can benefit from direct assignment.
rfratto
force-pushed
the
river-capsule-type-loss-fix
branch
from
199b9f8 to
f8d2640
Compare
mattdurham
approved these changes
Oct 14, 2022
Yeah, I'm a little nervous that I'm missing something and that something is going to break after this is merged, but we'll see what happens 🤷♂️ |
This commit changes the representation of objects and maps from
[]interface{} and map[string]interface{} to use []value.Value and
map[string]value.Value respectively.
This fixes grafana#2362, an issue where interface-based capsule values where
losing type information and being coerced to a type based on the actual
value.
This also greatly improves the performance of creating arrays and
objects, which no longer need to be inidividually copied in a loop:
```
name old time/op new time/op delta
Object/Non-capsule-10 1.84µs ± 8% 0.00µs ± 0% -99.89% (p=0.000 n=19+19)
Object/Capsule-10 1.82µs ± 4% 0.00µs ± 0% -99.89% (p=0.000 n=20+19)
Array/Non-capsule-10 389ns ± 2% 19ns ± 2% -95.20% (p=0.000 n=19+18)
Array/Capsule-10 389ns ± 1% 19ns ± 0% -95.19% (p=0.000 n=20+20)
name old alloc/op new alloc/op delta
Object/Non-capsule-10 1.88kB ± 0% 0.00kB -100.00% (p=0.000 n=20+20)
Object/Capsule-10 1.88kB ± 0% 0.00kB -100.00% (p=0.000 n=20+20)
Array/Non-capsule-10 344B ± 0% 24B ± 0% -93.02% (p=0.000 n=20+20)
Array/Capsule-10 344B ± 0% 24B ± 0% -93.02% (p=0.000 n=20+20)
name old allocs/op new allocs/op delta
Object/Non-capsule-10 42.0 ± 0% 0.0 -100.00% (p=0.000 n=20+20)
Object/Capsule-10 42.0 ± 0% 0.0 -100.00% (p=0.000 n=20+20)
Array/Non-capsule-10 2.00 ± 0% 1.00 ± 0% -50.00% (p=0.000 n=20+20)
Array/Capsule-10 2.00 ± 0% 1.00 ± 0% -50.00% (p=0.000 n=20+20)
```
As part of this change, the logic used for the concat function has been
greatly simplified. This has seemingly no impact on performance, as the
benchmark results below show:
```
name old time/op new time/op delta
Concat-10 1.10µs ± 4% 1.09µs ± 1% ~ (p=0.735 n=18+19)
name old alloc/op new alloc/op delta
Concat-10 1.30kB ± 0% 1.30kB ± 0% ~ (all equal)
name old allocs/op new allocs/op delta
Concat-10 20.0 ± 0% 20.0 ± 0% ~ (all equal)
```
rfratto
force-pushed
the
river-capsule-type-loss-fix
branch
from
f8d2640 to
c76d7ed
Compare
rfratto
added a commit
to rfratto/agent
that referenced
this pull request
Oct 26, 2022
… empty objects This change is similar to the change introduced in grafana#2369, where calling the `Interface()` method on River values forced a type change from an interface-based capsule to an any-type capsule. The new `value.FromRaw` method allows constructing a River value from a reflect.Value directly, avoiding the need to convert to interface{}. As part of this change, a `!reflect.Value.IsZero()` check has been added for safety; this shouldn't happen in most circumstances unless a value.Null is trying to be encoded. Fixes grafana#2361.
rfratto
added a commit
that referenced
this pull request
Oct 27, 2022
… empty objects (#2437) This change is similar to the change introduced in #2369, where calling the `Interface()` method on River values forced a type change from an interface-based capsule to an any-type capsule. The new `value.FromRaw` method allows constructing a River value from a reflect.Value directly, avoiding the need to convert to interface{}. As part of this change, a `!reflect.Value.IsZero()` check has been added for safety; this shouldn't happen in most circumstances unless a value.Null is trying to be encoded. Fixes #2361.
github-actions
bot
added
the
frozen-due-to-age
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit changes the representation of objects and maps from
[]interface{}andmap[string]interface{}to use[]value.Valueandmap[string]value.Valuerespectively.This fixes #2362, an issue where interface-based capsule values where losing type information and being coerced to a type based on the actual value backing the interface.
This also greatly improves the performance of creating arrays and objects, which no longer need to be inidividually copied in a loop:
As part of this change, the logic used for the concat function has been greatly simplified. This has seemingly no impact on performance, as the benchmark results below show:
I didn't update the CHANGELOG because #2362 only manifested for the
otelcolcomponents, which haven't been released yet.