flowchartsman

// package spooler implements a disk-persistent queue.
//
// Spooler uses MDB (LMDB) to implement a queue of byteslices. Its intended usecase
// is to enqueue work items received by a service before later working them off.
// Note that Spooler only flushes to disk up to once every 25ms. As a result,
// if the process or machine crashes unexpectedly, the most recent 25ms of work
// can be lost. This decision effectively increases throughput by 10,000%,
// but makes spooler unsuitable for jobs that absolutely cannot be allowed to fail
// under any circumstances.
package spooler

flowchartsman

[Unit]
Description=Internet Routing Registry daemon
After=network.target

[Service]
Type=simple
ExecStart=/usr/sbin/irrd -m

[Install]
WantedBy=multi-user.target

flowchartsman

#!/usr/bin/env python
"""
Takes snapshot of full webpages using Webkit and Qt4
usage:
python web2png.py URL [FILE]
example:
python web2png.py http://python.org out.png

see:
http://riverbankcomputing.com/static/Docs/PyQt4/html/qwebview.html#load

flowchartsman

#!/usr/bin/env bash

# Usage:
# 	./git_move.sh git@repo_site.com:/my_repo.git origin/folder/path/ /destination/repo/path/ new/folder/path/  

repo=$1
folder=$2
dest_repo=$3
dest_folder=$4
clone_folder='__git_clone_repo__'

flowchartsman

Locate the section for your github remote in the .git/config file. It looks like this:

[remote "origin"]
	fetch = +refs/heads/*:refs/remotes/origin/*
	url = git@github.com:joyent/node.git

Now add the line fetch = +refs/pull/*/head:refs/remotes/origin/pr/* to this section. Obviously, change the github url to match your project's URL. It ends up looking like this:

flowchartsman

Keybase proof

I hereby claim:

• I am alaska on github.
• I am alaska (https://keybase.io/alaska) on keybase.
• I have a public key ASA6ZY8rNwnzkpJ1KP4zWYQdmPXnNCRrV2dKMLzP3e2b3wo

To claim this, I am signing this object:

flowchartsman

Overview

This gist is based on the information available at golang/dep, only slightly more terse and annotated with a few notes and links primarily for my own personal benefit. It's public in case this information is helpful to anyone else as well.

I initially advocated Glide for my team and then, more recently, vndr. I've also taken the approach of exerting direct control over what goes into vendor/ in my Dockerfiles, and also work from isolated GOPATH environments on my system per project to ensure that dependencies are explicitly found under vendor/.

At the end of the day, vendoring (and committing vendor/) is about being in control of your dependencies and being able to achieve reproducible builds. While you can achieve this manually, things that are nice to have in a vendoring tool include:

flowchartsman

package main

import (
	"net/http"

	"github.com/gorilla/mux"
)

// example of a basic handler
func fooHandler(w http.ResponseWriter, r *http.Request) {

flowchartsman

package main

import (
	"net/http"

	"github.com/gorilla/mux"
)

// example of a basic handler
func fooHandler(w http.ResponseWriter, r *http.Request) {

flowchartsman

v, ok := inter.(MyType)
if !ok {
  return errors.New ("nope")
}

// becomes

switch v := inter.(type) {
case MyType:
	// do MyType stuff with v