Hi, I've looked into these 2-3 implementations of JSON decoders/encoders and found you are right in every word. Sorry for the possible confusion I've caused. On the other hand I'm worried about complexities which newcomers like me will meet :-)
Do you think that sample implementation in examples/docs/contrib will be good for dealing with this? At the same time you won't have to support requests such as mine for contrib stuff. Thanks, Timur On Fri, Jul 31, 2015 at 10:01 PM, Rob Miller <[email protected]> wrote: > On 07/31/2015 11:51 AM, Timur Batyrshin wrote: > >> Sounds good. Thanks for considering it! >> >> Do you think creating a mapping in the way similar to >> PayloadRegexDecoder would work here? >> > No, I don't. Describing such mappings using declarative configuration > tends to be quite awkward. I think a generic decoder should look for keys > that match the default message fields (i.e. "hostname", "payload", > "severity", etc.) and store those where you'd expect. Then all remaining > data would be written out to the dynamic message fields, with nested JSON > structures flattened into top level keys, such as what's happening here: > https://github.com/mozilla-services/lua_sandbox/blob/dev/modules/util.lua#L13 > > If you want to support a custom JSON to Heka message field mapping, then > just write your own SandboxDecoder. The Lua code to do this would be > shorter and easier to read than any declarative mapping syntax we might > define. > > -r > > >> >> Timur >> >> >> On 31 Jul 2015 20:58:22, Rob Miller <[email protected]> wrote: >> >> ------------------------------------------------------------------------ >> We don't currently have a generic JSON decoder. Mainly this is b/c >> generic JSON doesn't map directly to Heka message fields, so >> generally we encourage folks to spend a few minutes actually >> thinking about how they'd like their JSON input to be mapped to the >> Heka message structure. It would be possible, and is probably a good >> idea, however, to write one that does this for you automatically. It >> would have to recurse through the JSON structure, collapsing it into >> a flat structure that can be represented by Heka message fields. >> I've opened an issue for it: >> https://github.com/mozilla-services/heka/issues/1651 -r On >> 07/31/2015 08:51 AM, Timur Batyrshin wrote: > I've seen that. By >> "this module" I'm referring to encoder/decoder. > > On Fri, Jul 31, >> 2015 at 6:41 PM, Simon Pasquier >> >> >
_______________________________________________ Heka mailing list [email protected] https://mail.mozilla.org/listinfo/heka
