dmke

$ ruby -v
ruby 2.1.1p76 (2014-02-24 revision 45161) [x86_64-linux]

$ ruby re_bench.rb 
Rehearsal --------------------------------------------------------
(?-mix:^[a-zA-Z]$)     0.780000   0.000000   0.780000 (  0.783924)
(?i-mx:^[a-z]$)        0.790000   0.000000   0.790000 (  0.789410)
(?-mix:\A[a-zA-Z]\z)   0.370000   0.000000   0.370000 (  0.373894)
(?i-mx:\A[a-z]\z)      0.370000   0.000000   0.370000 (  0.373349)
----------------------------------------------- total: 2.310000sec

dmke

-# app/views/devise/registrations/edit.html.haml

%h1.page-header
  Edit
  = resource_name.to_s.humanize

= form_for resource, as: resource_name, url: registration_path(resource_name), html: { method: :put, class: 'form-horizontal' } do |f|
  = devise_error_messages!

  .form-group

dmke

random_twitter_message = (1..140).map{ rand 0xFF }.pack("C*")
hexdump random_twitter_message

# 0000:  af 98 02 2b 25 13 01 42  b8 cd 3a a5 23 07 c2 43  |...+%..B ..:.#..C|
# 0010:  c0 af bd 65 3e a7 c0 4c  26 b9 1e 54 cc c3 5b 3f  |...e>..L &..T..[?|
# 0020:  db d6 86 76 22 d4 31 ab  32 fe b6 32 f4 d4 1c 01  |...v".1. 2..2....|
# 0030:  f6 5d 25 8c 64 9b 18 3a  9d 50 14 a2 86 79 a6 f2  |.]%.d..: .P...y..|
# 0040:  da 50 f3 70 b6 bc e4 76  15 db 2f b7 f6 9c 88 d0  |.P.p...v ../.....|
# 0050:  63 95 23 66 b3 20 f6 ab  62 a6 28 92 c7 6d a9 e0  |c.#f. .. b.(..m..|
# 0060:  1c 2d fe 12 3e fd 8b 95  47 4d b2 14 07 e8 6b 6c  |.-..>... GM....kl|

dmke

Just a small benchmark to clear up the myth, Ruby wouldn'd garbage collect any Symbols. In fact, MRI 2.2.0 introduced Symbol GC in December 2014 (see notable changes in the release announcement).

The benchmark is quite simple: It basically cretes a bunch of Symbols via String#to_sym and counts Symbol::all_symbols afterwards.

I've used rbenv to test the Symbol GC (or the lack thereof) of these interpreter implementations:

• 1.9.3-p551
• 2.0.0-p645
• 2.1.6
• 2.2.2

dmke

influxdb-ruby: 0.3.0
influx server: 0.10.2

purging existing data...
generating data:
  UNIT_000 : {:consumed_energy=>5.0, :frequency=>5.0, :power_factor=>5.0, :reactive_watts=>5.0, :rms_voltage=>5.0, :temperature=>5.0}
  UNIT_001 : {:consumed_energy=>5.0, :frequency=>5.0, :power_factor=>5.0, :reactive_watts=>5.0, :rms_voltage=>5.0, :temperature=>5.0}
  UNIT_002 : {:consumed_energy=>5.0, :frequency=>5.0, :power_factor=>5.0, :reactive_watts=>5.0, :rms_voltage=>5.0, :temperature=>5.0}
  UNIT_003 : {:consumed_energy=>5.0, :frequency=>5.0, :power_factor=>5.0, :reactive_watts=>5.0, :rms_voltage=>5.0, :temperature=>5.0}
  UNIT_004 : {:consumed_energy=>5.0, :frequency=>5.0, :power_factor=>5.0, :reactive_watts=>5.0, :rms_voltage=>5.0, :temperature=>5.0}

dmke

#!/bin/bash -e

self=$(readlink -f "$0")
root=$(dirname "$self")

version="$1"
checksum="$2"

if [ -z "$version" ]; then
  echo >&2 "Missing version specifier."

dmke

#include zbt.dsti

&spi0 {
	m25p80@0 {
		partition@50000 {
			reg = <0x50000 0x1fb0000>
		};
	};
};

dmke

"Frozen soft reset" on MT7261 SoCs with 32M flash

This hack is derived from [this mail][] by Paul Fertser on openwrt-devel.

• When using 32M flash, the SPI bus needs to be reconfigured to use 4-byte addressing mode, instead of the usual (?) 3-byte mode.
• Issueing a soft reset without this patch leaves the bus in 4-byte mode, which then crashes the system because the SoC tries (and fails) to communicate in 3-byte addressing mode.
• Writing the opcodes brings the SPI back into 3-byte addressing mode,

dmke

package main

var n = func() int { return 42 }()

var (
	m = func() int { return 23 }()
	o = getTruth()
)

func main() {

dmke

Performing UDP tunneling through an SSH connection

Intro

The Swiss ISP [Bluewin][] sucks. Their DNS are often down. A friend even received advice from Bluewin technicians to not use their own DNS!... But then, it is quite hard to gain access to another DNS for free, if you don't have access to a co hosted machine.

In this document, we'll access another machine's network internal DNS services (UDP