Generally speaking, I am more interested in getting the high level use
cases right and being able to demonstrate to newbies how to solve
problems like the MovieLens examples. This will allow us to get more
users and contributors (Gremlin-Scala already has a number of
contributors).
Driving the development from that level suits me better because I learn
more about the kind of problems we can solve with Gremlin. Getting 100%
coverage of all the permutations of steps is nice (and I do maintain a
large number of the standard testsuite), but relatively far more effort
in comparison to what I achieve and learn. Here's what makes it hard
from my perspective:
* Failing tests are often quite hard to debug. I it throws a just submit
a query and get 'false was not true' back. A few times I ended up adding
random System.out.println in gremlin-test java code, then building that
artifact and running my tests again. Very time consuming process.
* you guys update your tests pretty often, which is good, but that means
a lot of effort for me, and I'm doing this in my spare time
* AFAIK it's all or nothing per suite - if a suite has 20 tests and one
fails for some unknown reason I can't just ignore that one. Maybe
there's a way to ignore tests, not sure
On 09/28/2015 11:42 PM, Stephen Mallette wrote:
Michael, can you please remind me of why there is "a lot of effort" in
using the test suite?
On Sun, Sep 27, 2015 at 4:59 PM, Michael Pollmeier <
[email protected]> wrote:
Painless. Note however that I don't implement the whole standard test
suite, as it's a lot of effort to take care of all the nitty gritty bits.
That's where the pain is from version to version, and I'm just being lazy
and avoid it.
Instead I focus on more high level examples, e.g. getting all steps from
your MovieLens presentation to work. Which works nicely and is fun.
Cheers
Michael
On 09/25/2015 05:11 PM, Marko Rodriguez wrote:
Hey Michael,
Can you tell us what your experience was bumping Gremlin-Scala to
TinkerPop 3.0.1?
Thanks,
Marko.
http://markorodriguez.com
