feat: Add valkey support by seedspirit · Pull Request #4 · instructkr/base-cacheable-class

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit · 2025-07-27T08:16:12Z

resolve issue #3 (Valkey Support)

Valkey Support

Added the valkey interface
Improved performance by using valkey-glide instead of valkey-py inside
However, we chose to inject a separately created config dataclass in consideration of the possibility that it will be replaced by valkey-py in the future.

Provided Decorator Factory for DX improvement

Currently, it is cumbersome to initialize separate CacheInterface and Cache Decorator each time, so we added a Factory class to shorten the process.

Test Code

Added a test framework for the valkey cache decorator
Refactored to inject decorator fixtures to prevent rewriting the same test code repeatedly for each decorator type.

Ilevk

here's some comments.

Ilevk · 2025-07-27T11:56:57Z

+
+ result = await func(*args, **kwargs)
+ if result is not None:
+ await self.cache.set(_key, result, ttl=current_ttl)


p3. even if result is None, I think it's better to cache. it helps the code not query a source repeatedly.

But redis / valkey don’t allow to set None value. so you should find other way.
In my case, I used a special value to represent None in valkey. e.g) '_NA'

Ilevk · 2025-07-27T12:08:28Z

+logger = logging.getLogger(__name__)
+
+
+class ValkeyCacheDecorator(CacheDecoratorInterface):


p3. I'll return to fundamental questions. do we need a decorator implementation for each cache?
I think the sync/async decorator is enough, adding another kind of cache would make the source code duplicate a lot.
for example, redis cache decorator and valkey cache decorator have a similar interface and logic. (mostly same).

Ilevk · 2025-07-27T12:10:23Z

+ cached_keys: list[str] = await self.cache.get_keys_regex(
+ target_func_name=target_func_name, pattern=pattern
+ )
+ for cache_key in cached_keys:


p3. you can use pipeline to send commands at once. but it might need to add interface for pipeline in cache class.

seedspirit · 2025-07-27T13:06:11Z

 dependencies = [
 "redis>=6.2.0",
+ "valkey-glide>=2.0.1",
 ]


Make dependency optional

seedspirit · 2025-07-27T13:08:29Z

+ async def _serialize(self, value: Any) -> bytes:
+ data = pickle.dumps(value)
+ return data
+
+ async def _deserialize(self, data: bytes | None) -> Any:
+ if data is None:
+ return None
+ data = pickle.loads(data) # noqa: S301
+ return data


I think we can serialize with mspack
Can we sperate serial/deserializer layer?

We need @sigridjineth's opinion

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

seedspirit added 2 commits July 27, 2025 16:34

feat: Implement valkey support
979204c

test: Add test codes for valkey cache decorator
b957f37

Ilevk reviewed Jul 27, 2025
View reviewed changes

seedspirit commented Jul 27, 2025
View reviewed changes

		logger = logging.getLogger(__name__)


		class ValkeyCacheDecorator(CacheDecoratorInterface):

Uh oh!

Conversation

seedspirit commented Jul 27, 2025

Uh oh!

Ilevk left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

IlevkJul 29, 2025• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

seedspirit commented Jul 27, 2025

Uh oh!

Ilevk left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

IlevkJul 29, 2025• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

seedspirit commented Jul 27, 2025

Uh oh!

Ilevk left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

IlevkJul 29, 2025•
edited
Loading

IlevkJul 29, 2025•
edited
Loading

IlevkJul 29, 2025•
edited
Loading