[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
[technology-pmc] Nomination of Exempt Pre-reqs for cloud software
|
Hi all,
As the PMC for LocationTech, we can identify various dependencies and
ask that they be considered 'exempt prerequisites' by the EMO. To frame
this discussion, check out
https://eclipse.org/org/documents/Eclipse_Policy_and_Procedure_for_3rd_Party_Dependencies_Final.pdf.
If I'm understanding that pdf correctly, an exempt pre-req enjoys the
following benefits:
1. Projects would not need to file CQs for any version of the pre-req.
2. Projects would be able to compile against new versions of an exempt
pre-reqs without further discussion, etc.
3. Projects could ship code from an exempt pre-req's project.
Just to be clear, I think #1 and #2 are pretty standard and clear. For
#3, I'm being aggressive in reading the document since it is silent and
a broad interpretation will help with at least one use case*.
Even without #3, I believe we'd get a nice benefit in not needing to
document and require review for the various versions of cloud software
in play.
As a first pass, I'd like to suggest that these libraries be considered
for recommendation to the EMO as exempt prerequisites. (This list is not
exhaustive, but it should reflect several of the cloud dependences from
the cloud-based LocationTech projects.)
1. Apache Hadoop (https://hadoop.apache.org/)
2. Apache Accumulo (https://accumulo.apache.org/)
3. Apache Zookeeper (https://zookeeper.apache.org/)
4. Apache Spark (https://spark.apache.org/)
For each of these projects, Apache, Cloudera, and Hortonworks all
provide slightly different versions to create consistent cloud
architectures in the same manner that a Linux distribution modifies and
assembles numerous open source projects to create a uniform product.
Use of GeoJinni/SpatialHadoop, GeoMesa, GeoTrellis, and GeoWave assume
such a cloud system.
We expect that versions of these packages to be installed, and reviewing
the numerous custom versions modified by Cloudera and Hortonworks would
be impractical.
* In terms of shipping code, I *think* Rob mentioned a bit of GeoTrellis
code which extended or required a bit of uncommonly deployed Spark code
to add some monitoring capabilities. If we can ship exempt pre-reqs,
that may help with that corner case.
Anyhow, in terms of procedure, I'd suggest we discuss 1) my broad
interpretation of the linked pdf, and 2) the initial list of suggested
exemptions.
Thanks in advance,
Jim