[Proposal] Use std-uritemplate instead of maintaining a separate implementation
#7
Conversation
| @@ -77,7 +77,7 @@ public void testSimpleStringExpansion() { | |||
| assertEquals("one=1,two=2,three=3,comma=%2C", UriTemplate.of("{map*}").expandToString(variables)); | |||
| assertEquals("", UriTemplate.of("{empty_map}").expandToString(variables)); | |||
| assertEquals("", UriTemplate.of("{empty_map*}").expandToString(variables)); | |||
| assertExpansionFailure("{list:1}"); | |||
There was a problem hiding this comment.
I belive this is a bug in the current implementation and the updated result is the expected one.
| } | ||
| return pos; | ||
| } | ||
| private List<String> extractTokens() { |
There was a problem hiding this comment.
The diff in GH looks pretty ugly, anyhow, to respect the current option to throw an exception on missing keys (which is a usability "extra" on top of the original spec), I implemented a single pass extraction of the tokens as an additional validation step.
|
cc. @cescoffier |
|
@cescoffier @vietj are you interested into(ever) going down this path, or do you want me to close the PR? |
|
Polite ping |
|
@andreaTP sorry missed the pR? we need to evaluate your library |
|
@vietj let me know if I can help. The C# implementation is the most used uri template library on NuGet: |
👋 Hi there, and thanks for the great project!
This is a proposal that I'd like to evaluate with this project maintainers.
In the Kiota project we faced the challenge of having to support implementations of uri-template in various languages, for this and more motivations we ended up rolling out a new one: std-uritemplate .
More recently I created a Quarkus-Kiota extension and alternative libraries, as expected, the provided Http Client implementation for the Quarkus integration is based on VertX 🙂 .
Now, having two independent implementations of the same (simple)algorithm on the classpath doesn't surely hurt much the users, but, merging the efforts has a couple of pretty interesting benefits:
With this PR I'm tentatively spearheading the collaboration opportunity, showing the tradeoffs of the approach in it's practical implementation.
Additional notes:
talking about performance, in the early days of
std-uritemplateI have been assisted by @franz1981 and I just updated a quick JMH benchmark to demonstrate that the two implementation have comparable results.Happy to hear your thoughts!