polygene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Merlin <paulmer...@apache.org>
Subject Re: [DISCUSSION] Ver 3.0 coming...
Date Fri, 07 Apr 2017 21:17:13 GMT
Hey,

Yay for the upcoming 3.0-RC1!


Le 2017-04-07 17:54, Niclas Hedhman a écrit :
> Hi,
> I want to push hard for a release at end of next week. I am looking 
> through
> the remaining 3.0 marked issues, and
> 
> Timeseries --> postpone

OK

> Entity = Identity+Value --> Postpone indefinitely (too much 
> consequence)

We can look at this one in very different ways and see small increments 
that can give us some goodness without eating the whole cake. What I 
would like us to do quickly, and preferably for 3.0, is to change how 
Entity state is serialized in our EntityStore SPI helpers.

Today, the "value" part of an EntityState is mixed with entity aspects:

{
   reference: "..",
   application_version: "..",
   type: "..",
   version: "..",
   modified: "..",
   properties: {..},
   associations: {..},
   manyassociations: {..},
   namedassociations: {..}
}

I'm thinking about persisting entities with the following structure 
instead:

{
   identity: "..",
   application_version: "..",
   type: "..",
   version: "..",
   modified: "..",
   value: { // Exact same state (de)serialization as values
     myProperty: {..},
     myAssoc: "..",
     myManyAssoc: [..],
     myNamedAssoc: {..}
   }
}

This is quite simple to change, simplifies a lot of code and I'm willing 
to push that forward rather quickly.
I already have a local experiment of this change.
BUT, this is a breaking change of the storage format so before doing so 
I'd like to know what do you all think.
3.0 is a good time to do this. Don't want to wait for 4.0. If we want to 
push a version out prior to that change, then it should be a -ALPHA1 
instead of a -RC1.

> Java 9 issues --> postpone, Java 9 is not out yet

OK, everything works without Jigsaw already
There's only the Riak client that does not support Java 9 yet.

> Yeoman generator --> will complete

If we release the generator to npm we need to sort out how to do that 
the ASF way.

> ORM --> just postponed, too much effort to release within weeks let 
> alone
> days.

OK

> Pluggable Types --> I don't know that status. Paul? Postpone if needed.
> (POLYGENE-94)

I answered directly on the issue. Please follow up there.

> Mapping in toEntity/toValue --> Will investigate, possibly implement
> otherwise postpone.

This should not be too complicated, but we can also postpone to 3.1 as 
an API enhancement.

> Rhino -> javax.scripting --> working on it.

Cool

> Jar Signing --> I don't know, and look at Paul the build master for
> opinion. (POLYGENE-116)

Postpone! It may be a bit of a can of worms.

> OSGi not working --> Postpone (I am secretly giving up on OSGi)

I personally don't use OSGi at all and don't care much.

> Cocnerns/SideEffect on methods --> I will investigate, test and 
> document if
> it works. Otherwise postpone.

OK

> POLYGENE-121 + 122 --> would like to fix, IF I knew better what you 
> meant.

Those are minor enhancements we can do in 3.1.

> POLYGENE-132 --> Might be a big one. I have not fully understand what 
> Kent
> was concluding, but might be important.

Kent?

> Indexing-SQL  --> Really tough one. I don't think I have enough time to 
> fix
> this. So, do we cut it out of the release and revive it later? Or is 
> there
> someone volunteering to take a stab at it?

Stan?




Mime
View raw message