Storage/eager ast #7125

johanfylling · 2024-10-18T15:06:44Z

Implements: #4147

Based on PoC by @ashutosh-narkar

Signed-off-by: Ashutosh Narkar <anarkar4387@gmail.com>

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

* Handle stored AST values in `bundle.ReadBundleEtagFromStore()` * Adding tests Signed-off-by: Johan Fylling <johan.dev@fylling.se>

ashutosh-narkar

The changes look fine @johanfylling. The test coverage is solid! Few comments 👇 . I had a high level comment about how we write data to the store. Currently this is decided by how the data is to be read. Did you consider always writing data as AST to the store and then returning based on the user-provided setting? In this way we don't have two paths in the store implementation.

Secondly, we should add some docs that highlight the pros and cons of this feature.

ashutosh-narkar · 2024-10-24T21:59:04Z

bundle/store.go

@@ -997,6 +1030,7 @@ func LegacyReadRevisionFromStore(ctx context.Context, store storage.Store, txn s

 // ActivateLegacy calls Activate for the bundles but will also write their manifest to the older unnamed store location.
 // Deprecated: Use Activate with named bundles instead.
+// FIXME: Test this with AST-read toggled on?


We should either address this or we could remove the legacy activation if we think enough time has passed.

Removing legacy activation feels like it should be it's own thing, tracked separately. I'll add some tests.

ashutosh-narkar · 2024-10-24T22:08:13Z

cmd/flags.go

@@ -165,6 +165,11 @@ func addV1CompatibleFlag(fs *pflag.FlagSet, v1Compatible *bool, value bool) {
 	fs.BoolVar(v1Compatible, "v1-compatible", value, "opt-in to OPA features and behaviors that are enabled by default in OPA v1.0")
 }

+func addReadAstValuesFromStoreFlag(fs *pflag.FlagSet, readAstValuesFromStore *bool, value bool) {
+	// FIXME: naming?


Seems like a good name to me.

The concern I have here is that without the right context, --read-ast-values tells you almost nothing about what this flag does.
Changing it to --read-ast-values-from-store would give some more context; but it's almost misleading, since reading AST values is more of a side effect of what we're actually doing: eagerly converting stored data to AST.

--read-ast-values-from-store seems better and if we add a good description that could help.

eagerly converting stored data to AST.

If we always stored AST then it wouldn't, no? But that has other consequences.

From a UX perspective, the naming here requires insight into OPA internals to understand. Many sysadmin types hardly know what an AST is, but they could know that some policies evaluate slowly and want to fix that.

Perhaps we can consider a name that reflects the effect (faster eval at the cost of higher memory usage) rather than on how that is accomplished? Could even leave room for future optimizations to that end that are unrelated to the AST handling at all.

The implementation details can of course still be covered in the docs for anyone curious.

How about something like --optimize-inmem-store-for-read-speed?
It's a bit of a mouthful. We could drop the inmem part, but this feature doesn't apply to disk stores (but this we can also clarify in the description). We could drop the read part; but writing, such as bundle updates, will actually be slower. We could swap speed for perf/performance, which most will probably equate to processing speed anyways; but it's a bit too broad, as initial memory footprint/perf will be worse (but likely more stable over time).

ashutosh-narkar · 2024-10-24T22:11:50Z

cmd/eval.go

+	v0Compatible           bool
+	v1Compatible           bool
+	traceVarValues         bool
+	ReadAstValuesFromStore bool


Maybe add some tests for the cli package for eval at least?

ashutosh-narkar · 2024-10-24T22:23:04Z

rego/rego.go

@@ -579,6 +579,7 @@ type Rego struct {
 	compiler                    *ast.Compiler
 	store                       storage.Store
 	ownStore                    bool
+	ownStoreReadAst             bool // FIXME: Alternative to a new option, we could add a WritableStore() method that assigns store and sets ownStore to true.


Is there any drawback of adding a new option?

Not really. having a WritableStore() option would give us some more flexibility, as the API user could then hand any store to the rego instance and let it write to it. But the current solution gives us a simple on-toggle that should be easier to understand. So I'm good with leaving this as-is.

ashutosh-narkar · 2024-10-24T22:24:14Z

runtime/runtime.go

@@ -240,6 +240,9 @@ type Params struct {

 	// CipherSuites specifies the list of enabled TLS 1.0–1.2 cipher suites
 	CipherSuites *[]uint16
+
+	// FIXME: Document this


ashutosh-narkar · 2024-10-24T23:47:37Z

storage/inmem/ast.go

+		obj.Foreach(func(k *ast.Term, v *ast.Term) {
+			if k.Equal(key) {
+				return
+			}
+			items = append(items, [2]*ast.Term{k, v})
+		})


This will be very expensive for a large object (and array below) but I don't think we have a way around it.

Yeah, it's unfortunate. Unless we find some other way of doing this, this is an up-front cost we need to eat on delta bundle updates. A cost that compounds with the eager json-to-AST conversion. If this turns out to be a too large cost, we might need to consider another approach completely; such as lazy conversion, but storing values for re-use (but that has its own drawbacks).

This would be good to document so that users understand the consequences of doing this.

ashutosh-narkar · 2024-10-24T23:57:34Z

storage/inmem/ast_test.go

+			expected: `{"a": 1, "b": 42, "c": 3}`,
+		},
+		{
+			// new keys can be added to objects


What the difference between this test case and set object key:

Good catch! This looks like a copy-paste mistake. Will update the test case to be an actual addition.

ashutosh-narkar · 2024-10-25T00:20:34Z

storage/inmem/inmem.go

+
+	// returnASTValuesOnRead, if true, means that the store will eagerly convert data to AST values,
+	// and return them on Read.
+	// FIXME: naming(?)


Seems ok to me.

Maybe. To reiterate my previous point. Since what we're actually doing is eagerly converting to AST on write, this is a bit of a misnomer. We could consider this to be an implementation detail, but then we're kinda hiding the actual purpose behind this feature, which is to eliminate conversion cost on read.

returnASTValuesOnRead is doing what it says so it seems like an implementation detail.

ashutosh-narkar · 2024-10-25T00:24:00Z

storage/inmem/inmem.go

@@ -327,11 +345,45 @@ func (h *handle) Unregister(_ context.Context, txn storage.Transaction) {
 }

 func (db *store) runOnCommitTriggers(ctx context.Context, txn storage.Transaction, event storage.TriggerEvent) {
+	if db.returnASTValuesOnRead && len(db.triggers) > 0 {
+		// FIXME: Not very performant for large data.


Maybe this a caveat we should point out about using this feature so users are aware of this. There seems to be no way to escape the conversion here.

Yeah .. an alternative would be to not do the conversion here, and push that responsibility to the consumer. Then conversion would only happen when the consumer is interested in the value. Or the consumer could even be updated to deal with AST values in addition to raw data. That'd be a breach in interface contract, though, and a pretty gnarly breaking change that is only detected at runtime, not compiletime.

Imo making the consequences more clear seems better that making a breaking change in this case.

We can always optimize.

ashutosh-narkar · 2024-10-25T00:36:44Z

storage/inmem/inmem.go

+	if s.returnASTValuesOnRead {
+		s.data = ast.NewObject()
+	} else {
+		s.data = map[string]interface{}{}
+	}
+


Interesting..we are controlling how data is written to the store based on returnASTValuesOnRead. Did you consider always writing AST values to the store and return AST/Go values based on returnASTValuesOnRead.

Always converting data to AST values would have a huge impact on OPA:s memory footprint and on bundle update performance. If these weren't concerns, then I think there is an argument for completely dropping the old raw data option altogether, as topdown is agnostic to the data type. Unfortunately, I think this feature is something only for users that either have very little data (at which point the performance boost is likely pretty small anyways), or have the ability to sacrifice these things to improve eval performance.

Having an option to do lazy conversion, but storing converted values for re-use (which I've mentioned in other places), might be a middle ground that might be enough of the best of both worlds that it could be enabled by default. This is however an unsubstantiated claim that I haven't thought too hard about.

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

netlify · 2024-10-25T15:06:39Z

✅ Deploy Preview for openpolicyagent ready!

Name	Link
🔨 Latest commit	`b515b9d`
🔍 Latest deploy log	https://app.netlify.com/sites/openpolicyagent/deploys/6721e414a919b20008e0462c
😎 Deploy Preview	https://deploy-preview-7125--openpolicyagent.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

To edit notification comments on pull requests, go to your Netlify site configuration.

ashutosh-narkar · 2024-10-25T22:38:34Z

@johanfylling thanks for addressing my comments. Do you have any performance numbers you can share that show the improvements and potential drawbacks of this change?

johanfylling · 2024-10-28T15:06:33Z

Here are some measurements against the test bundle we've used to evaluate this feature:

opa bench (disabled):

opa bench data.policy.ingress -b parameter-test-cases
+-------------------------------------------------+--------------+
| samples                                         |           69 |
| ns/op                                           |     16246874 |
| B/op                                            |     16602796 |
| allocs/op                                       |       428321 |
| histogram_timer_rego_external_resolve_ns_75%    |          333 |
| histogram_timer_rego_external_resolve_ns_90%    |          375 |
| histogram_timer_rego_external_resolve_ns_95%    |          437 |
| histogram_timer_rego_external_resolve_ns_99%    |         1250 |
| histogram_timer_rego_external_resolve_ns_99.9%  |         1250 |
| histogram_timer_rego_external_resolve_ns_99.99% |         1250 |
| histogram_timer_rego_external_resolve_ns_count  |         69.0 |
| histogram_timer_rego_external_resolve_ns_max    |         1250 |
| histogram_timer_rego_external_resolve_ns_mean   |          277 |
| histogram_timer_rego_external_resolve_ns_median |          250 |
| histogram_timer_rego_external_resolve_ns_min    |          125 |
| histogram_timer_rego_external_resolve_ns_stddev |          151 |
| histogram_timer_rego_query_eval_ns_75%          |     17299104 |
| histogram_timer_rego_query_eval_ns_90%          |     19534417 |
| histogram_timer_rego_query_eval_ns_95%          |     21096479 |
| histogram_timer_rego_query_eval_ns_99%          |     23926584 |
| histogram_timer_rego_query_eval_ns_99.9%        |     23926584 |
| histogram_timer_rego_query_eval_ns_99.99%       |     23926584 |
| histogram_timer_rego_query_eval_ns_count        |         69.0 |
| histogram_timer_rego_query_eval_ns_max          |     23926584 |
| histogram_timer_rego_query_eval_ns_mean         |     16228025 |
| histogram_timer_rego_query_eval_ns_median       |     15488667 |
| histogram_timer_rego_query_eval_ns_min          |     13838958 |
| histogram_timer_rego_query_eval_ns_stddev       |      2204325 |
+-------------------------------------------------+--------------+

opa eval --profile (disabled):

opa eval --profile --format=pretty --count=10 -b ./parameter-test-cases 'data.policy.ingress'
{
  "allow": true
}
+--------------------------------+-----------+-----------+-----------------+-------------------+----------------+
|             METRIC             |    MIN    |    MAX    |      MEAN       |        90%        |      99%       |
+--------------------------------+-----------+-----------+-----------------+-------------------+----------------+
| timer_rego_data_parse_ns       | 78743874  | 87138586  | 8.16318753e+07  | 8.67692151e+07    | 8.7138586e+07  |
| timer_rego_external_resolve_ns | 291       | 8583      | 1220.7          | 7799.700000000002 | 8583           |
| timer_rego_load_bundles_ns     | 112969917 | 125805667 | 1.171694626e+08 | 1.251271128e+08   | 1.25805667e+08 |
| timer_rego_module_compile_ns   | 31520875  | 37062375  | 3.35296167e+07  | 3.69148459e+07    | 3.7062375e+07  |
| timer_rego_module_parse_ns     | 6809627   | 11737961  | 9.0055879e+06   | 1.16811895e+07    | 1.1737961e+07  |
| timer_rego_query_compile_ns    | 54209     | 124209    | 77483.4         | 123667.3          | 124209         |
| timer_rego_query_eval_ns       | 14280000  | 17962292  | 1.67852792e+07  | 1.7932017e+07     | 1.7962292e+07  |
| timer_rego_query_parse_ns      | 35916     | 68041     | 46933.3         | 68007.8           | 68041          |
+--------------------------------+-----------+-----------+-----------------+-------------------+----------------+
+------------+-------------+-------------+-------------+-------------+----------+----------+--------------+---------------------------------------------------------------------------------------+
|    MIN     |     MAX     |    MEAN     |     90%     |     99%     | NUM EVAL | NUM REDO | NUM GEN EXPR |                                       LOCATION                                        |
+------------+-------------+-------------+-------------+-------------+----------+----------+--------------+---------------------------------------------------------------------------------------+
| 9.953458ms | 12.760292ms | 11.758979ms | 12.753292ms | 12.760292ms | 1        | 1        | 1            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:69 |
| 4.141582ms | 5.108708ms  | 4.829816ms  | 5.108495ms  | 5.108708ms  | 2        | 2        | 2            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:77 |
| 44.376µs   | 69.625µs    | 56.054µs    | 69.162µs    | 69.625µs    | 2        | 2        | 2            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:46 |
| 24.793µs   | 49.083µs    | 35.699µs    | 48.737µs    | 49.083µs    | 1        | 1        | 1            | data.policy.ingress                                                                   |
| 23.125µs   | 47.625µs    | 30.862µs    | 46.57µs     | 47.625µs    | 3        | 3        | 3            | parameter-test-cases/policy/ingress/test.rego:16                                      |
| 10.542µs   | 29.459µs    | 16.929µs    | 29.033µs    | 29.459µs    | 1        | 1        | 1            | parameter-test-cases/policy/ingress/test.rego:28                                      |
| 10.042µs   | 17.333µs    | 12.737µs    | 17.312µs    | 17.333µs    | 1        | 1        | 1            | parameter-test-cases/policy/ingress/test.rego:36                                      |
| 8.041µs    | 15.333µs    | 11.162µs    | 15.174µs    | 15.333µs    | 1        | 1        | 1            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:47 |
| 7.541µs    | 13.458µs    | 9.558µs     | 13.166µs    | 13.458µs    | 2        | 2        | 2            | parameter-test-cases/global/library/authorization/app_roles/app_roles.rego:26         |
| 5.625µs    | 10.833µs    | 7.591µs     | 10.633µs    | 10.833µs    | 1        | 0        | 1            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:35 |
+------------+-------------+-------------+-------------+-------------+----------+----------+--------------+---------------------------------------------------------------------------------------+

opa bench (enabled):

opa bench data.policy.ingress -b parameter-test-cases --read-ast-values
+-------------------------------------------------+--------------+
| samples                                         |        20536 |
| ns/op                                           |        55442 |
| B/op                                            |        49699 |
| allocs/op                                       |          857 |
| histogram_timer_rego_external_resolve_ns_75%    |          125 |
| histogram_timer_rego_external_resolve_ns_90%    |          126 |
| histogram_timer_rego_external_resolve_ns_95%    |          166 |
| histogram_timer_rego_external_resolve_ns_99%    |          209 |
| histogram_timer_rego_external_resolve_ns_99.9%  |         1281 |
| histogram_timer_rego_external_resolve_ns_99.99% |         1292 |
| histogram_timer_rego_external_resolve_ns_count  |        20536 |
| histogram_timer_rego_external_resolve_ns_max    |         1292 |
| histogram_timer_rego_external_resolve_ns_mean   |          106 |
| histogram_timer_rego_external_resolve_ns_median |         84.0 |
| histogram_timer_rego_external_resolve_ns_min    |         41.0 |
| histogram_timer_rego_external_resolve_ns_stddev |         62.3 |
| histogram_timer_rego_query_eval_ns_75%          |        51959 |
| histogram_timer_rego_query_eval_ns_90%          |        66138 |
| histogram_timer_rego_query_eval_ns_95%          |        74440 |
| histogram_timer_rego_query_eval_ns_99%          |        88304 |
| histogram_timer_rego_query_eval_ns_99.9%        |       126341 |
| histogram_timer_rego_query_eval_ns_99.99%       |       126542 |
| histogram_timer_rego_query_eval_ns_count        |        20536 |
| histogram_timer_rego_query_eval_ns_max          |       126542 |
| histogram_timer_rego_query_eval_ns_mean         |        49367 |
| histogram_timer_rego_query_eval_ns_median       |        44854 |
| histogram_timer_rego_query_eval_ns_min          |        38166 |
| histogram_timer_rego_query_eval_ns_stddev       |        11179 |
+-------------------------------------------------+--------------+

opa eval --profile (enabled):

opa eval --profile --format=pretty --count=10 -b ./parameter-test-cases --read-ast-values 'data.policy.ingress'
{
  "allow": true
}
+--------------------------------+-----------+-----------+-----------------+-----------------+----------------+
|             METRIC             |    MIN    |    MAX    |      MEAN       |       90%       |      99%       |
+--------------------------------+-----------+-----------+-----------------+-----------------+----------------+
| timer_rego_data_parse_ns       | 79425122  | 87189335  | 8.11994413e+07  | 8.6604589e+07   | 8.7189335e+07  |
| timer_rego_external_resolve_ns | 250       | 542       | 341.4           | 533.5           | 542            |
| timer_rego_load_bundles_ns     | 114970667 | 136677792 | 1.190086584e+08 | 1.348652586e+08 | 1.36677792e+08 |
| timer_rego_module_compile_ns   | 30945583  | 35549292  | 3.26016542e+07  | 3.55404711e+07  | 3.5549292e+07  |
| timer_rego_module_parse_ns     | 6777125   | 11474796  | 8.177038e+06    | 1.14064044e+07  | 1.1474796e+07  |
| timer_rego_query_compile_ns    | 49791     | 71542     | 61137.5         | 71037.8         | 71542          |
| timer_rego_query_eval_ns       | 149458    | 195916    | 169574.9        | 194978.6        | 195916         |
| timer_rego_query_parse_ns      | 35834     | 61583     | 43250           | 61474.7         | 61583          |
+--------------------------------+-----------+-----------+-----------------+-----------------+----------------+
+----------+----------+----------+----------+----------+----------+----------+--------------+---------------------------------------------------------------------------------------+
|   MIN    |   MAX    |   MEAN   |   90%    |   99%    | NUM EVAL | NUM REDO | NUM GEN EXPR |                                       LOCATION                                        |
+----------+----------+----------+----------+----------+----------+----------+--------------+---------------------------------------------------------------------------------------+
| 42.167µs | 62.499µs | 49.799µs | 62.198µs | 62.499µs | 2        | 2        | 2            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:46 |
| 20.332µs | 27.542µs | 23.937µs | 27.542µs | 27.542µs | 1        | 1        | 1            | data.policy.ingress                                                                   |
| 17.5µs   | 24.416µs | 20.653µs | 24.303µs | 24.416µs | 3        | 3        | 3            | parameter-test-cases/policy/ingress/test.rego:16                                      |
| 10.876µs | 22.167µs | 14.275µs | 21.75µs  | 22.167µs | 1        | 1        | 1            | parameter-test-cases/policy/ingress/test.rego:28                                      |
| 9.292µs  | 16.625µs | 12.15µs  | 16.604µs | 16.625µs | 2        | 2        | 2            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:77 |
| 8.333µs  | 11.375µs | 9.628µs  | 11.325µs | 11.375µs | 1        | 1        | 1            | parameter-test-cases/policy/ingress/test.rego:36                                      |
| 7.458µs  | 11.333µs | 8.749µs  | 11.241µs | 11.333µs | 2        | 2        | 2            | parameter-test-cases/global/library/authorization/app_roles/app_roles.rego:26         |
| 5.125µs  | 9.958µs  | 6.791µs  | 9.745µs  | 9.958µs  | 1        | 0        | 1            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:35 |
| 3.625µs  | 6.834µs  | 4.541µs  | 6.738µs  | 6.834µs  | 1        | 1        | 1            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:47 |
| 3.292µs  | 3.958µs  | 3.612µs  | 3.945µs  | 3.958µs  | 1        | 1        | 1            | parameter-test-cases/global/library/authorization/app_roles_internal/internal.rego:68 |
+----------+----------+----------+----------+----------+----------+----------+--------------+---------------------------------------------------------------------------------------+

I'll collect some measurements for bundle updates later (aiming for tomorrow).

ashutosh-narkar · 2024-10-28T17:12:31Z

bundle/store.go

@@ -59,9 +59,25 @@ func metadataPath(name string) storage.Path {
 	return append(BundlesBasePath, name, "manifest", "metadata")
 }

+func read(ctx context.Context, store storage.Store, txn storage.Transaction, path storage.Path) (interface{}, error) {


May be an optimization would be to avoid the conversions in this package by adding support for AST-based logic. This probably is a big enough change that can be worked on in the future if needed.

ashutosh-narkar

LGTM. @johanfylling if we add some documentation about this that would be great.

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

ashutosh-narkar

LGTM. Thanks for working on this @johanfylling!

ashutosh-narkar and others added 12 commits July 19, 2024 12:58

WIP: Store AST values in in-mem store

af9392c

Signed-off-by: Ashutosh Narkar <anarkar4387@gmail.com>

WIP: Commented out check make bankdata bundle compile

9c22743

Signed-off-by: Ashutosh Narkar <anarkar4387@gmail.com>

Merge branch 'main' into store-ast-val

bffb9e0

Make it possible to toggle between using raw- and AST data

845cd7a

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

Update AST hashes when mutating store data

4d8174b

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

Merge branch 'main' into storage/eager-ast

2012164

Adding common cmd flag

47a34ea

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

minor refactor

19a2411

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

Making linter happy

9cacd82

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

Merge branch 'main' into storage/eager-ast

b5bcdd1

Using consolidated store.data field for both raw and AST data

68bd7be

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

* Handle stored AST values in bundle.ReadBundleMetadataFromStore()

df72cb7

* Handle stored AST values in `bundle.ReadBundleEtagFromStore()` * Adding tests Signed-off-by: Johan Fylling <johan.dev@fylling.se>

johanfylling marked this pull request as ready for review October 21, 2024 15:35

ashutosh-narkar reviewed Oct 25, 2024

View reviewed changes

Making changes after review by @ashutosh-narkar

611895d

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

ashutosh-narkar reviewed Oct 28, 2024

View reviewed changes

ashutosh-narkar previously approved these changes Oct 28, 2024

View reviewed changes

Adding docs

25da332

Signed-off-by: Johan Fylling <johan.dev@fylling.se>

johanfylling dismissed ashutosh-narkar’s stale review via 25da332 October 29, 2024 15:06

ashutosh-narkar approved these changes Oct 29, 2024

View reviewed changes

Merge branch 'main' into storage/eager-ast

b515b9d

johanfylling merged commit 6af5e79 into open-policy-agent:main Oct 30, 2024
28 checks passed

johanfylling deleted the storage/eager-ast branch October 31, 2024 13:36

BrewTestBot mentioned this pull request Oct 31, 2024

opa 0.70.0 Homebrew/homebrew-core#196286

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Storage/eager ast #7125

Storage/eager ast #7125

johanfylling commented Oct 18, 2024

ashutosh-narkar left a comment

ashutosh-narkar Oct 24, 2024

johanfylling Oct 25, 2024

ashutosh-narkar Oct 24, 2024

johanfylling Oct 25, 2024

ashutosh-narkar Oct 25, 2024

anderseknert Oct 28, 2024

johanfylling Oct 29, 2024

ashutosh-narkar Oct 24, 2024

ashutosh-narkar Oct 24, 2024

johanfylling Oct 25, 2024

ashutosh-narkar Oct 24, 2024

ashutosh-narkar Oct 24, 2024

johanfylling Oct 25, 2024

ashutosh-narkar Oct 25, 2024

ashutosh-narkar Oct 24, 2024

johanfylling Oct 25, 2024

ashutosh-narkar Oct 25, 2024

johanfylling Oct 25, 2024

ashutosh-narkar Oct 25, 2024

ashutosh-narkar Oct 25, 2024

johanfylling Oct 25, 2024

ashutosh-narkar Oct 25, 2024

ashutosh-narkar Oct 25, 2024

ashutosh-narkar Oct 25, 2024

johanfylling Oct 25, 2024

johanfylling Oct 25, 2024

netlify bot commented Oct 25, 2024 •

edited

Loading

ashutosh-narkar commented Oct 25, 2024

johanfylling commented Oct 28, 2024

ashutosh-narkar Oct 28, 2024

ashutosh-narkar left a comment

ashutosh-narkar left a comment

Storage/eager ast #7125

Storage/eager ast #7125

Conversation

johanfylling commented Oct 18, 2024

ashutosh-narkar left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

netlify bot commented Oct 25, 2024 • edited Loading

✅ Deploy Preview for openpolicyagent ready!

ashutosh-narkar commented Oct 25, 2024

johanfylling commented Oct 28, 2024

Choose a reason for hiding this comment

ashutosh-narkar left a comment

Choose a reason for hiding this comment

ashutosh-narkar left a comment

Choose a reason for hiding this comment

netlify bot commented Oct 25, 2024 •

edited

Loading