More v2 changes #55
More v2 changes #55
Conversation
|
|
||
| task wrapper(type: Wrapper) { | ||
| gradleVersion = '2.0' | ||
| gradleVersion = '2.14-rc-3' |
There was a problem hiding this comment.
This is the version mentioned here: https://github.com/johnrengelman/shadow
| classifier = '' | ||
|
|
||
| // Shade all jars except for launchdarkly | ||
| relocate('com', 'com.launchdarkly.shaded.com') { |
There was a problem hiding this comment.
The potential problem with shading Gson is that we return a Gson type (JsonElement) from our public interface: https://github.com/launchdarkly/java-client/blob/fa61ceb5f343c53db01f88b417a623575ab9fc58/src/main/java/com/launchdarkly/client/LDClient.java#L298-L298
This means that if a user's code already imports Gson this JsonElement is a different type than they might be expecting.
There was a problem hiding this comment.
Hmm. I hadn't considered that when I made the suggestion to shade Gson. I think we should not shade it then-- leaking a shaded class into an external interface isn't a good practice.
In addition to just having the wrong type, interacting with the JsonElement is going to be hard-- you'll have to manipulate the shaded API, which exposes more shaded footprint.
There was a problem hiding this comment.
It might be appropriate to reconsider this approach. In my project, SBT (sbt-assembly) got confused by launchdarkly-client depending on com.google.code.gson:gson, but both artifacts included com/google/gson/Gson.class etc.
| relocate('okhttp3', 'com.launchdarkly.shaded.okhttp3') | ||
| relocate('okio', 'com.launchdarkly.shaded.okio') | ||
| relocate('org', 'com.launchdarkly.shaded.org') { | ||
| exclude("org.slf4j.*") |
There was a problem hiding this comment.
Good catch on excluding the logging facade.
|
👍 |
add more flag evaluation unit tests
More v2 changes. See comments in code