MIgrate away from Kotlin Documentable #10660
Conversation
| @@ -28,6 +28,12 @@ type JMap[K, V] = java.util.Map[K, V] | |||
| type JHashMap[K, V] = java.util.HashMap[K, V] | |||
| type JMapEntry[K, V] = java.util.Map.Entry[K, V] | |||
|
|
|||
| private val emptyListInst = JList() | |||
There was a problem hiding this comment.
Can we preemptively use Collections.emptyList here? I'd prefer to avoid issues due to some code modifying our "empty" list.
| private val emptyListInst = JList() | ||
| def JNil[A] = emptyListInst.asInstanceOf[JList[A]] | ||
|
|
||
| private val emptyMapInst = JMap() |
| ) | ||
|
|
||
| case class TypeParameter( | ||
| variance: "" | "+" | "-", |
There was a problem hiding this comment.
Why we decide to keep variance explicitly, when bounds are aggregated into Signature?
There was a problem hiding this comment.
variance goes before name and in some cases we would like to refer only to name without variance
|
|
||
| val ext = c.get(ClasslikeExtension) | ||
| override def pageForClasslike(c: DClasslike): ClasslikePageNode = ??? |
There was a problem hiding this comment.
Maybe we should add more explanatory message why calling this method fails than "Not Implemented"
| else Kind.Class(Nil, Nil) | ||
|
|
||
| private def kindForClasslike(classDef: ClassDef): Kind = | ||
| def typeArgs = classDef.getTypeParams.map(mkTypeArgument) |
| symbol: Symbol, | ||
| member: MemberExtension, | ||
| compositeExt: CompositeMemberExtension = CompositeMemberExtension.empty): Member = | ||
| new DClass( |
There was a problem hiding this comment.
I know we will discuss whether dokka will still remain as a base for our doctool, but if dokka stays, we should consider making Member a subtype od Documentable, and get rid of these hacks with DClass wrapper :P
There was a problem hiding this comment.
Actually using DClass is really conviniet :)
romanowski
force-pushed
the
scala3doc/member-migration
branch
from
fbb52ce to
a7ba826
Compare
romanowski
force-pushed
the
scala3doc/member-migration
branch
2 times, most recently
from
3a4a125 to
bb41d4f
Compare
Scala3doc was only loosly using dokka model and it has started to blocking us. This PR standarize over Member and expand semantic of Kind. This allow us to have more typesafe way to reason about code. It will also make migration away from dokka easier if needed.
…r fixes to external location mechanism.
romanowski
force-pushed
the
scala3doc/member-migration
branch
from
bb41d4f to
bcaa1a7
Compare
| @@ -46,182 +46,163 @@ class ScalaSignatureProvider(contentConverter: CommentsToContentConverter)(using | |||
|
|
|||
| private def stylesIfDeprecated(m: Member): Set[Style] = | |||
| if m.deprecated.isDefined then styles ++ Set(TextStyle.Strikethrough) else styles | |||
|
|
|||
|
|
|||
|
|
|||
| object ScalaSignatureProvider: | |||
| def rawSignature(documentable: Documentable, builder: SignatureBuilder): SignatureBuilder = | |||
There was a problem hiding this comment.
I will get rid of all Documentable after this PR is merged
Scala3doc was only loosly using dokka model and it has started to blocking us.
This PR standarize over Member and expand semantic of Kind. This allow us to have more typesafe way to reason about code.
It will also make migration away from dokka easier if needed.
This PR contains #10702 that allow to link to external APIs as well as refactor linking.
Beside that this PR document correctly all members as well as allows to link to given ones.
Signature are rendered correctly (no problem with annotations etc.)