Unexpected zinc cache invalidation after sbt remote cache hit #966
Comments
|
I have compared the relevant part of I still don't understand why though. Could it be related to https://github.com/AdoptOpenJDK/openjdk-jdk11/blob/19fb8f93c59dfd791f62d41f332db9e306bc1422/src/java.base/share/classes/java/lang/Object.java#L557 ? A classpath change between 1614849670371 & 1614849586071 that would either expose the annotation or not? The same JDK was used for both compilations. The fact that the - internal {
+ external {
key: "tv.teads.domains.advertiser.commands.ConnectTransactor"
value {
- compilationTimestamp: 1614849670371
+ compilationTimestamp: 1614849586071
name: "tv.teads.domains.advertiser.commands.ConnectTransactor"
api {
classApi {
name: "tv.teads.domains.advertiser.commands.ConnectTransactor"
access {
public {
}
}
modifiers {
flags: 1
}
selfType {
emptyType {
}
}
structure {
parents {
projection {
id: "Object"
prefix {
singleton {
path {
components {
id {
id: "java"
}
}
components {
id {
id: "lang"
}
}
components {
this {
}
}
}
}
}
}
}
parents {
projection {
id: "Any"
prefix {
singleton {
path {
components {
id {
id: "scala"
}
}
components {
this {
}
}
}
}
}
}
}
}
savedAnnotations: "scala.native"
savedAnnotations: "jdk.internal.HotSpotIntrinsicCandidate"
topLevel: true
}
objectApi {
name: "tv.teads.domains.advertiser.commands.ConnectTransactor"
access {
public {
}
}
modifiers {
}
definitionType: MODULE
selfType {
emptyType {
}
}
structure {
}
topLevel: true
}
}
- apiHash: 176768709
+ apiHash: -488308903
nameHashes {
name: "notify"
hash: 1514289040
}
nameHashes {
name: "wait"
hash: 1963066972
}
nameHashes {
name: "equals"
hash: 1626438048
}
nameHashes {
name: "asInstanceOf"
hash: -2008440760
}
nameHashes {
name: "synchronized"
hash: 1581358093
}
nameHashes {
name: "notifyAll"
hash: 287659096
}
nameHashes {
name: "isInstanceOf"
hash: -42141924
}
nameHashes {
name: "ConnectTransactor"
hash: 1351354997
}
nameHashes {
name: "=="
hash: 1290070777
}
nameHashes {
name: "clone"
hash: -1872329449
}
nameHashes {
name: "toString"
hash: -799817619
}
nameHashes {
name: "!="
hash: -1478124343
}
nameHashes {
name: "getClass"
hash: 1196203508
}
nameHashes {
name: "ne"
hash: -1383471943
}
nameHashes {
name: "eq"
hash: -491545741
}
nameHashes {
name: "##"
hash: 1393487132
}
nameHashes {
name: "finalize"
- hash: -245298853
+ hash: 1383774555
}
nameHashes {
name: "hashCode"
hash: -2114247454
}
- extraHash: 176768709
+ extraHash: -488308903
}
} |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
steps
Unfortunately, I haven't been able to extract a repro for this, as this issue is transient (but consistent for a given remote cache artifact). I decided to open a ticket to start adding information along my investigation to see if I can find a pattern.
Context
problem
Zinc runs incrementally after a remote cache artifact hit on
IntegrationTest, detecting a change in an untouched marker traittrait ConnectTransactordefined inCompile(on which Zinc had a cache hit after pulling the remote cache).expectation
Zinc cache hit - "No changes"
notes
Here are
incOptions.value.withApiDebug(true)logs:I have tried to diff the text format representation of
inc_compile.binforCompile&IntegrationTestbefore and after the recompilation, but I haven't found anything interesting (the diff is hard to read as order changes). I am not including it here as it's massive, but happy to provide extracts.The text was updated successfully, but these errors were encountered: