marklit

\set PROMPT1 '%[%033[1;33m%]%M:%>%[%033[1;32m%] %n@%/%R%[%033[0m%]%# '
\set PROMPT2 ' %R > '
\x auto
\timing
\set HISTFILE ~/.psql_history- :DBNAME
\set HISTCONTROL ignoredups
\set COMP_KEYWORD_CASE upper
\pset null '¤'

marklit

// See: https://imgur.com/gafDVW5

package com.example.mark.myapplication;

import android.os.Bundle;
import android.support.design.widget.FloatingActionButton;
import android.support.design.widget.Snackbar;
import android.support.v7.app.AppCompatActivity;
import android.support.v7.widget.Toolbar;
import android.view.View;

marklit

// See: https://imgur.com/6o6DP4Q

package com.example.mark.myapplication;

import android.os.Bundle;
import android.support.design.widget.FloatingActionButton;
import android.support.design.widget.Snackbar;
import android.support.v7.app.AppCompatActivity;
import android.support.v7.widget.Toolbar;
import android.view.View;

marklit

-- From: http://michael.otacoo.com/postgresql-2/postgres-9-5-feature-highlight-brin-indexes/
CREATE TABLE brin_example AS SELECT generate_series(1,100000000) AS id;
CREATE INDEX btree_index ON brin_example(id);

-- By default, the block size is 128 pages. This resolution can be increased or decreased using the pages_per_range
CREATE INDEX brin_index ON brin_example USING brin(id);
CREATE INDEX brin_index_64 ON brin_example USING brin(id) WITH (pages_per_range = 64);
CREATE INDEX brin_index_256 ON brin_example USING brin(id) WITH (pages_per_range = 256);
CREATE INDEX brin_index_512 ON brin_example USING brin(id) WITH (pages_per_range = 512);
-- The lower the pages per block, the more space the index will occupy, but the less lossy the index will be, i.e. it will need to discard fewer rows.

marklit

trip_id                 integer
vendor_id               string
pickup_datetime         TIMESTAMP
dropoff_datetime        TIMESTAMP
store_and_fwd_flag      string
rate_code_id            integer
pickup_longitude        float
pickup_latitude         float
dropoff_longitude       float
dropoff_latitude        float

marklit

Query 1

SELECT cab_type,
       count(*)
FROM trips
GROUP BY cab_type

marklit

$ (echo  -e "PRAGMA synchronous=OFF; PRAGMA count_changes=OFF;\n.mode csv\n.separator ,\n.import /dev/stdin trips";
   gunzip -c ~/taxi-data/*.csv.gz) | \
       sqlite3 trips.db

marklit

See GitHub for the source code and TensorFlow on a GTX 1080 for installation notes.

$ python fizz_buzz.py \
    --hidden_units 500 \
    --iterations 4000 \
    --learning_rate 0.04

marklit

- Inexplicable perversity of human nature.
- The clever machinations of MongoDB's marketing people.
- The AGPL license killed it.
- We spent too long development before monetizing.
- Bad performance.
- Numeric types limited to a 64-bit `float`.
- Great product, but didn't/couldn't translate to revenue.
- Bad business model.
- Failure in timezones/timestamp nuances.

marklit

git clone https://github.com/Smerity/cc-mrjob.git ~/cc-mrjob && cd ~/cc-mrjob
pip install -r requirements.txt
pip install awscli
aws configure

curl -O https://commoncrawl.s3.amazonaws.com/crawl-data/CC-MAIN-2017-04/wat.paths.gz
gunzip wat.paths.gz
python server_analysis.py \
    -r emr \
    --conf-path mrjob.conf \