[Interop-dev] Python packages

[Nemesis]

On 05/22/2015 12:28 PM, Mitar wrote:
> Hi!
>
>> I'm also interested in ujson. Which improvements are there? Will you try
>> to merge them back in the official package?
> Pull request is open:
>
> https://github.com/esnme/ultrajson/pull/157
>
> :-)

Very good stuff!

> So, from PostgreSQL you can get raw GeoJSON. So there is no use in
> parsing it first in Django into Python, and then serializing it again.
> To speed up GeoJSON REST API, I patched ujson so that it can read
> GeoJSON fields directly from PostgreSQL and then just dump it into REST
> output. So no expensive GDAL conversions are needed.
>
> https://github.com/wlanslovenija/nodewatcher/blob/development/nodewatcher/core/frontend/api/fields.py#L92
> https://github.com/wlanslovenija/django-datastream/blob/master/django_datastream/serializers.py#L36

Sounds interesting.

What about input? I guess this option is valid for output only.

When geojson data is received you still need to parse it, validate it,
and then if everything is ok put it into the database.

>> Many common functions (min, max, average, standard deviation, ecc.) are
>> built into the SQL-like langauge:
>> http://influxdb.com/docs/v0.8/api/aggregate_functions.html
> We have all that already working in datastream based on MongoDB. Maybe
> you should try it out. :-) And it supports not just numeric values, but
> also nominal and graph:
>
> https://datastream.readthedocs.org/en/latest/usage.html#types
>
> The reason is because we want a time-series of network topology through
> time. :-)

Cool!
Well I am not so keen on switching to mongodb or datastream yet, I will
keep my eye on it and in case influxdb doesn't play well I'll switch.

As said, I tried many options already, and inflxudb was the one I felt
most productive with.

Federico