As title suggests, we are hoping to get rid of support for Scala 2.10
in near future.
Reasons for this are -- as far as I understand, Christopher and Nate
can elaborate -- mostly related to significant problems 2.10 has with
respect to introspection of properties, so that there are many things
that just can not be done with scala 2.10 but that are doable with
2.11 and above.
This is somewhat similar to having stopped supporting Scala 2.9 with
Jackson 2.4 (2.3 still supported it).
One question we have, however, is how soon to do that. Main choices,
in my opinion, are:
- Drop from the next minor update, Jackson 2.9. This would have the
benefit of letting Scala module authors work to fix many fundamental
issues as early as possible (as in: now -- master is for
2.9.0-SNAPSHOT)
- Drop from next major update, 3.0. Since the current thinking is that
2.9 is the last minor update for 2.x (or at least significant minor
version), it would mean waiting for one more minor version, and
additional timespan of 6-8 months for final release
* note: we are likely to branch `master` off to 3.0 much sooner,
maybe in 2 months, so doing this would only delay dev start by that
amount.
Now: I think the decision should be based on question of how much
active development there is by projects that rely on Jackson Scala
2.10 support. I stress word "active", because simple usage of a stable
Jackson version is not affected: any project that uses any older
Jackson version (up to 2.8) would be fine.
Looking at 2 sources I have access to, it would seem that total usage
is actually quite evenly split between scala 2.10 and 2.11:
- MvnRepository lists 159 dependants for scala 2.10 module; 157 for 2.11
- Maven Central repo has similar download counts: 2.10 has 95k and 2.11 72k
o D/l counts for 2.10 have not changed much during past 12 months
o D/l counts for 2.11 have grown from 25k to 72k over past 12 months
So, from simple deployments one could say these versions are equally important.
But looking at MC repo's version info, it is easy to see that 2.10
usage is more of legacy kind:
* 80% of 2.10 downloads are for Jackson version 2.4 or below (2.4.4
itself is close to half of all usage)
where 2.11 is used with newer versions:
* 70% of 2.11 downloads are for Jackson version 2.5 or later (and 20%
for.. version 2.4.4)
This suggests that adoption of newer versions is strongly skewed
towards scala 2.11 module, and that maintaining Jackson 2.8 with full
support may well be reasonable support mechanism.
With this, I propose dropping of Scala 2.10 support from Jackson 2.9.
What say you?
-+ Tatu +-
ps. Starting this discussion on dev list first: if we reach consensus
of "early" dropping of support, with Jackson 2.9, I will next consult
User community as well. If we decide to keep support there isn't much
to announce.
--
You received this message because you are subscribed to the Google Groups
"jackson-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
