Adrian Cole
Adrian Cole
Brave (who's binary form in cassandra tracing context) uses the former (keep the two halves of the ID next to eachother in byte order). I think this makes most sense,...
So I was chatting with @ellispritchard who maintains https://github.com/Financial-Times/tapper, and @fishcakez who is using it in a mixed environment with finagle. Also, I've thought about this a while now. I...
Thx for the start. Widening X-B3-TraceId seems the most pragmatic start. For binary, we could consider the check length approach of growing thrift later or adopt census (aka dapper2) format...
Hi. There is no working tracer library in finagle except scribe. If there was http then folks who don't depend on non finagle libs (such as zipkin-finagle) can operate without...
> > @adriancole would changing the name to > "finagle-zipkin-scribe" and adding some documentation fit what you're > looking for? It seems like unnecessary work for us to add our...
PS what you mentioned would be an improvement as breaking the artifact ID can alert folks they should be looking elsewhere. I do think people will still use scribe anyway,...
one clarification: if finagle-zipkin could become the right thing (ie package http transport to let people switch over) this would be ideal. If there's only one working transport upstream, even...
> > @adriancole it looks like in the short > term, we can rename finagle-zipkin to finagle-zipkin-scribe, and we'll > update our docs to refer to zipkin-finagle. > ok and...
> > I think we'd be open to a community contributed finagle-zipkin-http if we > had a commitment from the community to help support it, although it seems a >...