Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Untangle external reader code #776

Open
wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Untangle external reader code #776

wants to merge 2 commits into from

Conversation

odenix
Copy link
Contributor

@odenix odenix commented Nov 4, 2024

See commit messages for details. Second commit is a pure rename refactoring.

import org.pkl.core.evaluatorSettings.PklEvaluatorSettings
import org.pkl.core.externalreader.ModuleReaderSpec
import org.pkl.core.externalreader.ReaderProcess
import org.pkl.core.externalreader.ResourceReaderSpec
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Given that the resolver types are shared between the client readers and external readers, is org.pkl.core.externalreader really the right place for these types?

Copy link
Contributor Author

@odenix odenix Nov 4, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

org.pkl.server depending on externalreader seems acceptable to me. These types couldn't stay in module/resource because that creates package cycles and is a layering violation: module/resource already export packagereader.ReaderProcess, hence packagereader shouldn't also depend on them.

pkl-core has serious modularity issues. I just tried to do some damage control in the few hours I had. I think I made a step in the right direction; perhaps you guys will find further improvements.

@bioball
Copy link
Contributor

bioball commented Nov 4, 2024

Thanks for the PR, and appreciate the help straightening this out!

FYI--this is a little late to get into the 0.27 release. However, we can get this into the main branch once this release is out.

We can also take some time to better look at the API of pkl-core and have a clear definition of what's public and what's not (and perhaps use JPMS).

@odenix
Copy link
Contributor Author

odenix commented Nov 4, 2024

FYI--this is a little late to get into the 0.27 release.

Just keep in mind that this PR makes breaking changes to APIs introduced in 0.27.

We can also take some time to better look at the API of pkl-core and have a clear definition of what's public and what's not (and perhaps use JPMS).

I'd appreciate this. It will likely take quite a while until pkl-core is ready for JPMS. But we can take one step at a time: define which packages are public, enforce at build time that they don't export private packages, etc.

@odenix
Copy link
Contributor Author

odenix commented Nov 4, 2024

For the next release, could we create a release branch at least one week before the release, and merge only bugfixes from then on?

@bioball
Copy link
Contributor

bioball commented Nov 4, 2024

Yeah, we should definitely add some more process here, including cut-off for features. We'll figure out what that should look like after we get 0.27 out and into the world.

Apologies for the lack of structure right now.

Copy link
Contributor

@bioball bioball left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Most of these changes look good! Although, I think let's undo some of the name changes here.

import org.pkl.core.util.Nullable;

/** An external process that reads Pkl modules and resources. */
public interface ExternalReaderProcess extends AutoCloseable {
public interface ReaderProcess extends AutoCloseable {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is a public API as of 0.27, so, let's just keep the name as-is.

Copy link
Contributor Author

@odenix odenix Nov 11, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I assume you mean just this interface? (There are quite a few breaking changes in this PR.)

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yup, the name of this interface!

return new Proxy(addressUri, noProxy);
}

@SuppressWarnings("unchecked")
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Either undo this change, or re-add and deprecate.

But, I think it's fine to just keep this method around; we don't gain much by inlining this.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The motivation is to remove a method from the public API that doesn't belong there. I can add it back as deprecated.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sorry, I don't fully understand; what makes Proxy.create() not belong here?

Copy link
Contributor Author

@odenix odenix Nov 13, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Sorry, my impression was that Proxy.create() is an internal method used by Proxy.parse() that accidentally became a public API.

I feel that regular Java code should just use new Proxy(new URI(...)) or new Proxy(URI.create(...)). These will also throw more appropriate exceptions than PklException, which is mainly thrown by the evaluator and doesn't seem ideal for regular Java APIs that don't evaluate Pkl code.

One more thing I noticed is that similar classes such as Release and PklInfo use java.lang.String for pkl.base.Uri. (They also don't have their own package.) A potential problem with using java.net.URI for pkl.base.Uri is that valid Pkl objects can't be represented in Java.

String scheme();
public interface ModuleResolver {
static ModuleResolver of(MessageTransport transport, long evaluatorId) {
return new ModuleResolverImpl(transport, evaluatorId);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There aren't any layering violations here, are there?

Since we've now shipped this in 0.27, I think let's keep this as-is, and have org.pkl.core.externalreader.ModuleReaderSpec implement interface Spec here.

Copy link
Contributor Author

@odenix odenix Nov 11, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ModuleReaderSpec is a simple record, and callers of ModuleKeys.externalResolver may need to instantiate it.
I think the ModuleResolver.Spec interface has no value (at least not anymore) and just obfuscates the public API.

This PR already makes breaking API changes. If it's important to make this particular change less breaking, record ModuleReaderSpec could be renamed to ModuleResolver.Spec. However, I doubt this trade of breaking changes has much benefit.

I prefer the name ModuleReaderSpec over ModuleResolver.Spec because it makes more sense in the ReaderProcess interface:

interface ReaderProcess {
  // used to return ModuleResolver.Spec, which I found confusing
  ModuleReaderSpec getModuleReaderSpec(String scheme) throws IOException;

  // used to return ResourceResolver.Spec, which I found confusing
  ResourceReaderSpec getResourceReaderSpec(String scheme) throws IOException;
}

String scheme();
public interface ResourceResolver {
static ResourceResolver of(MessageTransport transport, long evaluatorId) {
return new ResourceResolverImpl(transport, evaluatorId);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same comment; let's keep this class as-is, and have ResourceReaderSpec implement this interface

- move the following classes into package externalreader:
  - ExternalModuleResolver
  - ExternalResourceResolver
  - MessageTransportModuleResolver (renamed to ExternalModuleResolverImpl, made package-private)
  - MessageTransportResourceResolver (renamed to ExternalResourceResolverImpl, made package-private)
- replace interface ExternalModuleResolver.Spec with record ExternalModuleReaderSpec
- replace interface ExternalResourceResolver.Spec with record ExternalResourceReaderSpec
- translate between messaging.ResourceReaderSpec and ExternalResourceReaderSpec (eliminates dependency from messaging on higher layer)
- translate between messaging.ResourceResolverSpec and ExternalResourceResolverSpec (eliminates dependency from messaging on higher layer)
- add ServerMessages.ExternalReader and translate between this message component and the PklEvaluatorSettings.ExternalReader API
- add ServerMessages.Proxy and translate between this message component and the PklEvaluatorSettings.Proxy API
- change type of CreateEvaluatorRequest.allowedModules/allowedResources from List<Pattern>? to List<String>?
  - removes a lot of code
  - should not need to create a Pattern object to send a message
- remove public method evaluatorSettings.PklEvaluatorSettings.Proxy.create()
  - only seems useful internally, inlined
- class names were too long/repetitive
- package name provides sufficient context
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants