Julian,
thank you for that email. I think we all like very much tone
and content - this is very encouraging!
Yes, the DataWorks Summit in Berlin seem to fit very well. Just
discussed with Patrick -- not too much of an exercise to get an
abstract ready in due time.
Best,
Bernard.
On 12/7/2017 11:09, Julian Hyde wrote:
Thanks for responding, everyone.
There’s one really important point I’d like to make about incubation (and
Apache in general) that many people find counter-intuitive. There are 3 basic
things to work on: (a) the code, (b) the incubation tasks (e.g. trademark
search), and (c) the community.
By far the most important thing is the community. Apache folks often cite the
mantra “community over code”; this means that if you build a healthy community,
the code will look after itself.
Consider one of the most important “tasks” of incubation, namely producing
releases. The contributors often focus on the code, giving themselves a very
high bar in terms of the number of features to implement and bugs to fix before
producing a release. But it’s much better to just get a release out there,
warts and all. The process of producing the release (testing it, writing the
doc, promoting it) pulls the community together. People will discover those
“warts”, contribute fixes, and you will have your first new committers.
The first incubator release always takes WAY longer than you expect, and not
for the reason you expect. It takes a lot of effort to assemble the release
into an acceptable format, checking the licenses of dependencies, including the
necessary LICENSE and NOTICE files, and so forth. I recommend that you start
work on the first release very soon, and resist the temptation to put lots of
features into it.
If you want to build community (i.e. attract people who don’t work for IBM or
live in Zurich) promotion is essential. An active twitter account, blog posts,
and talks at conferences or meet ups where your potential users are in
attendance. (For example, DataWorks Summit Berlin[1] is in April and CFP ends
in one week. A lot of attendees would be interested in Crail, even at this
early stage.)
As for tasks, they are listed on the status page [2]. We can burn them down and
update the page over the next couple of months.
Julian
[1] https://dataworkssummit.com/berlin-2018/
<https://dataworkssummit.com/berlin-2018/>
[2] http://incubator.apache.org/projects/crail.html
<http://incubator.apache.org/projects/crail.html>
On Dec 6, 2017, at 7:19 AM, bernard metzler <[email protected]> wrote:
Hi all,
I'm Bernard Metzler from IBM Zurich Research Lab. For quite some
years now, my main interests are in design, implementation and
deployment of flexible and highly efficient I/O stacks. I worked
on specification and implementation of protocols and programming
interfaces for Remote Direct Memory Access (RDMA) and non-volatile
memory integration. I am representing IBM at the Board of the
Open Fabrics Alliance. My contributions to the open source community
include a communication subsystem for non-volatile memory
integration with the BlueGene supercomputer, and a software
only RDMA driver for Linux, which aims at enabling RDMA
applications at any host system w/o dedicated RDMA hardware.
I am in the process of submitting this driver to Linux upstream.
Ultimate goal is to enable RDMA applications (like Crail!) within
any cloud environment.
I worked in the context of several international research projects,
including the Human Brain Project, and the Square Kilometer Array
Project.
I am involved with Crail from it's very beginning, so far mainly
contributing to the design discussion. It is my first Apache project.
Cheers,
Bernard.
