Pavel, we cannot omit them. Because if JSON will be transformed back to Java and some java field has not null default value that will lead to not correct de-serialization in this case.
On Thu, Jun 16, 2016 at 4:54 PM, Pavel Tupitsyn <ptupit...@gridgain.com> wrote: > Why do we even write null fields? Can we just omit them? > This will be faster, more compact, and less ambiguous. > > Pavel. > > On Thu, Jun 16, 2016 at 12:51 PM, Denis Magda <dma...@gridgain.com> wrote: > >> This looks strange. Definitely for consistency reasons it makes sense >> that all ‘null’ objects including string are serialized as “null”. >> >> In regards to the compatibility stuff I think it can be resolved somehow. >> >> — >> Denis >> >> On Jun 16, 2016, at 12:15 PM, Alexey Kuznetsov <akuznet...@gridgain.com> >> wrote: >> >> Hi All! >> >> I'm working on migrating from outdated json-lib to Jackson issue >> IGNITE-3277 [1] and found that previous library serialize null strings as >> "" and null objects as "null", for example: >> {"a": "", "b": null} >> >> How about to serialize all values as "null" ? >> In this case we would definitely know that in Java before serialization >> to JSON was really NULL, not empty string. >> >> But this may break compatibility in some cases. >> >> Thoughts? >> >> >> 1. https://issues.apache.org/jira/browse/IGNITE-3277 >> >> -- >> Alexey Kuznetsov >> GridGain Systems >> www.gridgain.com >> >> >> > -- Alexey Kuznetsov GridGain Systems www.gridgain.com