Setup LaTeX

First of all a fully working LaTeX Environment is necessary. There are tons of tutorials to handle the installation.

• http://www.tug.org/mactex/ (I've used this)
• http://www.tug.org/texlive/

The local build automation process requires more tools to work:

• https://mg.readthedocs.io/latexmk.html‌‌  used to build the pdf version
• make (preinstalled on most unix systems)‌‌for windows: http://gnuwin32.sourceforge.net/packages/make.htm

I've used pandoc to generate other file formats based on my LaTeX report, like ms word or open document.

• https://pandoc.org/installing.html

Setup VSCode

The full installation of LaTeX ships two different applications to work with LaTeX files (TeXworks or TeXShop). These are really powerfull and extensive editors. But I'm more famliar with development environments like IntelliJ or VS Code. For this reason I prefered to use VS Code.  The following tutorial describes the vs code setup using the plugin LaTeX Workshop:

https://medium.com/@rcpassos/writing-latex-documents-in-visual-studio-code-with-latex-workshop-d9af6a6b2815

Now the document bootstrap will be created. I need a report and a presentation based on the same content and references. Each chapter and section will be stored in separated files.

.
├── chapters
│   └── introduction
│       ├── abstract.tex
│       └── references.tex
├── figures
│   ├── images
│   ├── pdf
│   └── tables
├── keynote.tex
├── report.tex
└── shared-deps.tex

Dependencies

All shared dependencies are stored in shared-deps.tex file. The file will be imported by report.tex and keynote.tex. A working file can be downloaded on github.

% ----- Deps -- Figures and Graphics
\usepackage{graphicx} % https://ctan.org/pkg/graphicx
\usepackage{tikz} % https://ctan.org/pkg/pgf

All packages can be found on the Comprehensive TeX Archive Network. If your code does not fully relay on the LaTeX default installation you have to customize your deployment container.

Setup Document

Your report.tex file contains your main document definition. You can import each chapter file to get a better overview of the structure.

\documentclass[a4paper, 12pt]{report}
\input{shared-deps.tex} % Load shared dependencies

\begin{document}

    \begin{titlepage}
        % your titlepage
    \end{titlepage}
    
    \chapter{Intro}
    \input{chapters/intro/intro}
    
\end{document}

Add References

You can use different references systems. I've worked with biblatex and its own data backend called biber .  Put all your references to a the bibliography.bib file. I've used BibDesk to manage my reference library.

@book{book:maintainable-software,
    Author = {Joost Visser, Sylvan Rigal},
    Isbn = {9781491953495},
    Keywords = {testing, benchmarking},
    Publisher = {O'Reilly Media},
    Title = {Building Maintainable Software - Java Edition},
    Year = {2016}}

Before you can use them, you have to import the required packages:

\usepackage[german]{babel} % Local language support
\usepackage[backend=biber,style=alphabetic]{biblatex}
\usepackage[hidelinks,bookmarks=true]{hyperref}
\addbibresource{bibliography.bib} % import glossary entries

Use references

Now you can use these entries in your report for example with the footnote command: \footnote{This is the content of footnote}. You can also use citiations in footnote with a custom style. Overwrite the \footcite command with  your stylings. Place the code (from stackexchange) in your report.tex befor your document begins.

\DeclareCiteCommand{\footcite}[\mkbibfootnote] 
	{\usebibmacro{prenote}}                                 
	{\usebibmacro{citeindex}
		\printfield{labelalpha}
		\setunit{\labelnamepunct}
		\printfield[citetitle]{title}
	}
  {\addsemicolon\space}
  {\usebibmacro{postnote}}

Now you'll be able to use your layout with a key from your bibliography in your footnotes:

\footcite{book:maintainable-software}

Print references

Print a list of your used references with the \printbibliography command in your report file where you want to place the page.

Add Glossaries

For glossaries and acronyms i've used the glossaries package. With this package you're able to store your entries in a separate file. In this file you define your glossary and acronym items as followed in the glossary.tex file:

\newacronym{go}{Go}{Golang}
\newglossaryentry{mutex}{
	name=Mutex,
	plural=Mutex,
	description={Englische Abkürzung für \textbf{m}utual \textbf{e}xclusion}
}

You can use this entries by importing the file in your report. Your glossary will be generated by the \makeglossaries command. To use these index in a automates pipeline with latexmk you have to add a .latexmk configuration. Follow this post to understand why.

\usepackage[acronym, toc]{glossaries} % https://ctan.org/pkg/glossaries

\makeglossaries
\loadglsentries{glossary}

Link your glossary and reference entries with the hyperref package.

If you run the report file in visual studio code you have to generate the index manually with the command: makeglossaries report .

Now you can print your glossarie tables with these commands:

\printglossary
\printglossary[type=\acronymtype, title=Begriffe und Abkürzungen, toctitle=Begriffe und Abkürzungen] 

Build Process

The report can now be built to a single pdf file. To simplify this process, use a Makefile with following content. Run make to build the document. Read the docs of the make tool to enhance the file.

# dependencies
# - gcc
# - latexmk

.PHONY: setup clean pub

REPORT_NAME = report
PRESENATION_NAME = keynote
OUT_DIR = dist/

all: clean setup report pub

setup:
    mkdir -p ${OUT_DIR}
    
report: report.tex
    latexmk -bibtex -diagnostics -jobname=${REPORT_NAME} --file-line-error --interaction=nonstopmode -pdf report.tex
   
keynote: presentation.tex
		latexmk -bibtex -diagnostics -jobname=${PRESENATION_NAME} --file-line-error --interaction=nonstopmode -pdf keynote.tex

pub:
    cp ${REPORT_NAME}.pdf ${OUT_DIR}
    cp ${PRESENATION_NAME}.pdf ${OUT_DIR}
    
clean: 
    latexmk -CA
    rm -rf dist
    rm -f ${REPORT_NAME}.pdf
    rm -f ${PRESENATION_NAME}.pdf

I've used a continuous build pipeline to archive my built reports.

Github Workflows

You can use for example github workflows. It's well integrated and easy to use. Create the file .github/workflows/build.yml and add following pipeline definition. I've used the action from xu-cheng. When the built process is finished, all files from dist folder will be available as zip archive to download.

name: build
on: [push, pull_request]

jobs:
    build_report:
        runs-on: ubuntu-latest
        name: Build Report
        steps:
          - name: Set up Git repository
            uses: actions/checkout@v2
          - name: Setup LaTeX
            uses: xu-cheng/latex-action@v2
            with:
              root_file: report.tex
          - uses: actions/upload-artifact@v2
            with:
              name: report
              path: report.pdf

CircleCI

You can also use the circleCI Service. The pipelines runs faster with this provider than github workflows. The free credits (2500) from circleci allows you to build the report about 70 (4m build time) to 140 (2m build time) times a week. Create a file named .circleci/config.yml and add following snippet. I've used the docker container from  aergus.

version: 2
jobs:
  build:
    docker:
      - image: aergus/latex
    steps:
      - checkout
      - run:
          name: Build Case Study
          command: make all
      - store_artifacts:
          path: dist/
          destination: dist

Dependency Management

A large topic was the handling of dependencies mainly in the context of internal networks. The prerequisite for the presented solutions are Go modules which are activated by default as of Go 1.13 (when there is a go.mod file available in the project root).

Florin Pățan from Jetbrains demonstrated the use of Go modules very well using the development environment GoLand. He also wrote a blog post about it.

Several solutions were presented to load dependencies via a proxy. In some cases they even work in a supplementary way.

• https://proxy.golang.org
• https://github.com/gomods/athens
• https://jfrog.com/integration/go-registry

Benchmarking

Many of the speakers have presented a number of different ways to debug their applications and identify performance problems.

There are packages which can be used directly from go subrepos to benchmark different ranges.

golang.org/x/benchmarks/build
Build is a benchmark that examines compiler and linker performance.

golang.org/x/benchmarks/driver
Package driver provides common benchmarking logic shared between benchmarks.

golang.org/x/benchmarks/garbage
Garbage is a benchmark that stresses garbage collector.

golang.org/x/benchmarks/http
HTTP is a benchmark that examines client/​server http performance.

golang.org/x/benchmarks/json
JSON benchmark marshals and unmarshals ~2MB json string with a tree-like object hierarchy, in 4*GOMAXPROCS goroutines.

Profiling data can be exposed by an endpoint with the http pprof package. This is very useful to visualize these data directly in the browser or use a cli interface like gom.

Debugging

In some cases you want to know a trace of all calls in your runtime. The runtime package from go provides these capabilities. An interesting article from Dave Cheney explains the details here.

With expvar you are able to extract debug informations from your production (like pprof for performance issues) environment over an endpoint.

A very interesting talk by Fabio Falzoi. A deep insight into garbage collection of Go. He has his slides on github as pdf available.

Webservices

If you run several goroutines in parallel, you can cause conflicts. Roberto Clapis explained how to find the causes and minimize them. To do this, he introduced his debugging methods and helped to understand the problem of contention better. Get the slides here.

Bill Kennedy has introduced a starter-kit for webservices. The focus is on robustness and configuration.

Functional Programming with Go

Last but not least, I would like to mention Eleanor McHugh's talk. She presented exciting implementations to apply functional concepts in Go.

lxc launch ubuntu:16.04 gitlab
lxc config device add gitlab backup-share disk path=/data source=/data/gitlab

Host Setup

To access the gitlab instance with ssh, you have to set a static ip to your container.

Open your lxd_dnsmasq.conf file or create it if you don't have one.

sudo vim /etc/default/lxd_dnsmasq.conf
dhcp-host=gitlab,10.0.10.2

Then restart your lxd bridge to apply the config.

sudo service lxd-bridge stop && sudo service lxd-bridge start

Port forwarding for ssh access to the container host:2222 -> 10.0.10.2:22

sudo iptables -t nat -A PREROUTING -i eth0 -p tcp --dport 2222 -j DNAT --to-destination 10.0.10.2:22

Cleaning ACL on the backup shared directory /your/backup/

setfacl -R -b /data/gitlab

Create git user on the host to map with git user in the container

groupadd -g 100998 git
useradd -u 100998 -g 100998 git

Setting ACL on the hosts backup directory

setfacl -R -m u:user:100998:rwX,g:100998:rwX /data/gitlab

Container Setup

Open a shell on your gitlab container

lxc exec gitlab bash

Install required packages

apt-get install curl openssh-server ca-certificates postfix

Add the gitlab repositories to your dependencies

curl -sS https://packages.gitlab.com/install/repositories/gitlab/gitlab-ce/script.deb.sh | bash

Intall and reconfigure gitlab. Don't forget to configure the backup path of your gitlab.

apt-get install gitlab-ce
gitlab-ctl reconfigure

Edit cronjobs for your backup automation with crontab -e

Setup backup crontab [every day at 1am]

0 1 * * * /opt/gitlab/bin/gitlab-rake gitlab:backup:create CRON=1

Sources

• https://linuxcontainers.org/lxd/getting-started-cli/

List available Ubuntu Images

lxc image list ubuntu:

List available CentOS Images

lxc image list images | grep centos

Container

Create CentOS based container

lxc launch images:centos/7/amd64 conainer-name

Create Ubuntu based container and start

lxc launch ubuntu:16.04 conainer-name

Creating a Ubuntu container without starting it

lxc init ubuntu:16.04 conainer-name

List all Conainter

lxc list

Display Container Infos

lxc info conainer-name

Start / Stop

lxc start conainer-name

Bash Access to Container

lxc exec container-name bash

Devices

Show Devices

lxc config device show conainer-name

Add Device

lxc config device add container-name device-name disk path=/container/path source=/host/path

Remove Device

lxc config device remove conainer-name device-name

Sources

• https://linuxcontainers.org/lxd/getting-started-cli/
• https://blog.ubuntu.com/2016/03/22/lxd-2-0-your-first-lxd-container