Hi Yacine, On 7 November 2014 10:56, Yacine Kheddache <yacine.khedda...@alyseo.com> wrote: > Please check this out and let us know what you are thinking about it : > https://github.com/Alyseo/zfs/tree/json
I took a look at the code. I like that the entire JSON object appears on one line, in keeping with the popular pattern of "linefeed-delimited JSON" used by tools like bunyan[1]. I think the output would be better without the needless whitespace between keys, values, colons, etc. I think there's a deeper problem with this relatively simplistic approach to generating JSON: the encoding of non-printing characters, or characters beyond the 7-bit ASCII space. You really need to escape both key and value strings, and possibly deal with locales, etc. The safest subset of UTF-8 for use here is 7-bit ASCII, with numeric escape sequences for characters beyond that in the Basic Multilingual Plane. I know Robert mentioned to you earlier that we already have code in libnvpair, and thus available to OpenZFS in general, to take an nvlist_t and emit it as JSON. This code does deal (at least minimally) with the escaping of strings: https://github.com/joyent/illumos-joyent/blob/master/usr/src/lib/libnvpair/nvpair_json.c#L29-L109 I like the idea in general, we just need to agree on a correct and well-specified output format (and implementation) if we are introducing new Public flags to zpool(1M) and zfs(1M) -- especially as this output is for programmatic consumption. Thanks. [1]: https://github.com/trentm/node-bunyan -- Joshua M. Clulow UNIX Admin/Developer http://blog.sysmgr.org _______________________________________________ developer mailing list developer@open-zfs.org http://lists.open-zfs.org/mailman/listinfo/developer
