Pavel, Makes sense, I think this is the best way.
-Val On Wed, Jan 25, 2017 at 12:27 AM, Pavel Tupitsyn <ptupit...@apache.org> wrote: > Val, > > Not sure about UUID. > It is a very common thing and writing it as an object will introduce a lot > of overhead (17 bytes now vs 40+ bytes for object). > > May be we should even add a special case for IgniteUuid? > > On Wed, Jan 25, 2017 at 1:52 AM, Dmitriy Setrakyan <dsetrak...@apache.org> > wrote: > > > Why not add IgniteUuid to BinaryContext.BINARYLIZABLE_SYS_CLSS? > > > > On Tue, Jan 24, 2017 at 10:55 AM, Valentin Kulichenko < > > valentin.kuliche...@gmail.com> wrote: > > > > > I actually think that both UUID and IgniteUuid should be plain > > serializable > > > classes, I don't see any reason for special processing for them. > > > > > > Currently we have the following: > > > > > > - UUID is Serializable, but we have special serialization logic for > it > > > internally in the marshaller. > > > - IgniteUuid is Externalizable. > > > > > > This is indeed inconsistent and confusing. > > > > > > -Val > > > > > > On Tue, Jan 24, 2017 at 9:21 AM, Pavel Tupitsyn <ptupit...@apache.org> > > > wrote: > > > > > > > Igniters, > > > > > > > > Currently IgniteUuid is written with OptimizedMarshaller > > > > (it is not included in BinaryContext.BINARYLIZABLE_SYS_CLSS). > > > > > > > > This prevents it from being read on other platforms (.NET, C++). > > > > > > > > Is there any reason for this? Can we fix this in 2.0? > > > > > > > > Thanks, > > > > Pavel > > > > > > > > > >
