Indeed - it's a good one - so good that someone's taken it up! (There
is a UCR student - not talkative on this list - who's been working on
this one, and whom Steven is helping.)
On 2/1/17 7:58 AM, Sattam Alsubaiee wrote:
I might be obsolete :-), but one thing that I think worth looking at is
exploring and implementing other effecient merge policies for LSM indexes.
The good thing is that it is a self-contained problem and can be done
easily in 2-3 months.
Cheers,
Sattam
On Wed, Feb 1, 2017 at 9:16 AM, Yingyi Bu <[email protected]> wrote:
can have them create automated unit tests for many of our components
(Our Http Server comes to mind).
My $0.02: writing white box unit tests (e.g., MockIto-based stuff) are
devs' own responsibility..
But I guess that there could be two projects along the line:
1. re-organize our current runtime tests, e.g., avoiding unnecessary
DDL/DMLs.
2. cleanup sporadic failures in integration tests, particularly, managix
start/stop etc.
Best,
Yingyi
On Tue, Jan 31, 2017 at 10:01 PM, abdullah alamoudi <[email protected]>
wrote:
Taewoo,
I think this one would be too much for a Google Summer of Code student.
I am not sure if we can do this but if we can have a software testing
project, then we can have them create automated unit tests for many of
our
components (Our Http Server comes to mind). That would be very helpful I
think but definitely not very exciting.
Cheers,
Abdullah.
On Jan 31, 2017, at 9:56 PM, Taewoo Kim <[email protected]> wrote:
I would like to suggest one feature: adding a security feature to the
AsterixDB: setting DB accounts and privileges. Also, use them when
access
the Web Interface or API call.
Best,
Taewoo
On Tue, Jan 31, 2017 at 9:52 PM, Yingyi Bu <[email protected]> wrote:
Yeah, sounds good to me.
Best,
Yingyi
On Tue, Jan 31, 2017 at 9:31 PM, Steven Jacobs <[email protected]>
wrote:
Makes sense, based on a past of not quite finishing. There has been
some
discussion from the Summer of Code Apache threads, and it sounds like
they
want to do a smaller number of better vetted projects this year, but
I
think we could still try for at least one spot.
Steven
On Tue, Jan 31, 2017 at 9:21 PM Yingyi Bu <[email protected]>
wrote:
IMO, each of them could be a two-three month project.
The project itself could be conceptually simple and has a relatively
small
scope, but we probably should lift the exit bar.
Here are my thoughts for the exit criteria:
1. the feature reliably works for all edge cases;
2. the implementation does not hack things, which can result in
general-purpose refactoring of other modules;
3. there are regression unit tests and integration tests along with
the
code;
4. the code is carefully reviewed by 1-2 committers.
Getting sth. done as a proof-of-concept and then disappearing
doesn't
seem
a very effective approach..:-)
Best,
Yingyi
On Tue, Jan 31, 2017 at 9:07 PM, Steven Jacobs <[email protected]>
wrote:
Do you think these could be a single project (dataset
improvements)?
I'm
wondering about whether one alone would be a full two-three month
project.
Steven
On Tue, Jan 31, 2017 at 9:00 PM Yingyi Bu <[email protected]>
wrote:
I have two proposals for data loaders:
1. add AWS S3 as an external data source;
2. support POSTing a user file on the client side through HTTP
into
an
internal dataset.
Thanks!
Best,
Yingyi
On Thu, Jan 19, 2017 at 2:06 PM, Steven Jacobs <[email protected]>
wrote:
Hi All,
The Google Summer of Code official start email was sent out
today.
This
can
be a useful opportunity for us as an Apache software. We will
probably
want
to look at potential Asterix projects to suggest for student
proposals.
Steven
