How to setup the workflow of documentation build for your project ================================================================== **Setup you repository and then clone locally** ssh-add your-ssh-key git clone ssh://@gerrit.opnfv.org:29418/ **Inside the repository create the following structure:** gerrit.opnfv.org/ - docs/ --- docs/main.rst --- docs/other-docus.rst --- docs/images/*.png|*.jpg **In order to obtain a nice .html & .pdf at then end you must write you documentation using reSt markup** a quick guide: http://docutils.sourceforge.net/docs/user/rst/quickref.html **Clone the releng repository so you can created jobs for JJB** git clone ssh://@gerrit.opnfv.org:29418/releng cd releng/jjb// Create build-docu.sh with the following content: ------------------------------------------------- ``#!/bin/bash set -xv for file in $(find . -type f -iname '*.rst'); do file_cut="${{file%.*}}" html_file=$file_cut".html" pdf_file=$file_cut".pdf" rst2html $file > $html_file rst2pdf $file -o $pdf_file done`` Edit .yml and make sure you have the job-templates set as in the example below: ``- job-template: name: 'opnfvdocs-daily-{stream}' builders: - shell: !include-raw build-docu.sh - shell: | gsutil cp docs/*.pdf gs://artifacts.opnfv.org/opnfvdocs/docs/ gsutil cp docs/*.html gs://artifacts.opnfv.org/opnfvdocs/docs/ - job-template: name: 'opnfvdocs-verify' builders: - shell: !include-raw build-docu.sh - job-template: name: 'opnfvdocs-merge' builders: - shell: !include-raw build-docu.sh - shell: | gsutil cp docs/*.pdf gs://artifacts.opnfv.org/opnfvdocs/docs/ gsutil cp docs/*.html gs://artifacts.opnfv.org/opnfvdocs/docs/`` git add build-docu.sh .yml git commit --signoff #add the proper message to commit git review -v **Create the documentation using the recommended structure in your repository and submit to gerrit for review** **Jenkins will take over and produce artifacts in the form of .html & .pdf** Jenkins has the proper packages installed in order to produce the artifacts. **Artifacts are stored on Google Storage (still to decide where, structure and how to present them)** **NOTE:** In order to generate html & pdf documentation the needed packages are rst2pdf & python-docutils if the Jenkins is CentOS/RHEL; many variants have been tested but this is the cleanest as a solution. **Other options:** 1. Doxygen plugin -> HTML published plugin (html)/ LaTeX (pdf) --------------------------------------------------------------- Description: - html: using Doxygen plugin + HTML publisher It involves some customization at doxygen level + custom html header/footer - pdf: it generates a .pdf using latex Final destination of generated files needs to be discussed as it will be part of a Bash script in Post-actions. Input files: .md , .rst Output: .html & .pdf Pros: - standard tools: doxygen, html publisher, LaTeX suite - doxygen plugin available in Jenkins, you just need to install it; html publisher plugin available in Jenkins, you just need to install it - destination files are generated fast - standard reStructuredText or Markdown Cons: - takes some time to customize the output in matters of template, requires custom html header/footer - latex suite is quite substantial in amount of packages and consumed space (around 1.2 GB) Tested: roughly, functional tbeeingests only 2. Maven & clouddocs-maven-plugin (actually used to generate openstack-manuals) -------------------------------------------------------------------------------- Description: It represents the standard tool to generate Openstack documentation manuals, uses maven, maven plugins, clouddocs-maven-plugins; location of finally generated files is the object of a small Bash script that will reside as Post-actions Input files: .xml Output: .html & .pdf Pros: - quite easy for initial setup - uses openstack documentation generation flows as for openstack-manuals (clouddocs-maven-plugin), maven installs all you need generate the documentation Cons: - could be tricky to generate a custom layout, knowledge about Maven plugins required, .pom editing - dependent of multiple maven plugins - input files are .xml and xml editing knowledge is required Tested: roughly, functional tests only 3. Sphinx & LaTeX suite ------------------------ Description: The easiest to install, the cleanest in matter of folder & files structure, uses standard tools available in repositories; location of finally generated files is the object of a small Bash script that will reside as Post-actions Input files: .rst as default Output: .html & .pdf Pros: - standard tools: Python Sphinx, LaTeX suite - destination files are generated fast - standard reStructuredText as default; other inputs can be configured - Sphinx's installation is very clean in matters of folder structure; the cleanest from all tested variants - latex suite is also easy to install via yum/apt and available in general repos - everyone is migration from other tools to Spinx lately; it provides more control and better looking documentation - can be used also for source-code documentation, specially if you use Python Cons: - takes some time to customize the output in matters of template, requires custom html header/footer - latex suite is quite substantial in amount of packages and consumed space (around 1.2 GB) Tested: roughly, functional tests only