By Ricardo Fusco on 2017-05-26 12:39 in GSoC 17 PR testing Platform

Time: 24/05/2017,  15:00 UTC

Meeting Topics:

  1. Attempts with Jenkins so far

  2. Discussion about the build systems used for tests

  3. Discussion about Github webhooks and Github API

  4. Discussion about swarm and first scripts

Meeting Attended by:

Ashan Fernando, Niels Braczek, Tito Alvarez, Yves Hoppe, Ricardo Fusco, Tobias Zulauf

Summary

  1. Attempts with Jenkins so far

The first tests with Jenkins were not successful due to problems setting up the webhook  and also due to problems in the Github Pull Requests Builder plugin and Jenkins version, had to retry installing Jenkins and the necessary plugins, next tests will be in that sense so that Jenkins can be correctly configured to build the PRs and commits. Also missing a tool to test webhooks locally, planning on using Ngrok, a tool that creates a tunnel for incoming requests so that the webhooks can be tested in a local context.  

  1. Discussion about the build systems used for tests

The build systems used in Joomla! CMS for the different tests were mentioned in order to figure out how to insert the jenkins job to build PRs/commits in between or at the end of the jobs from the other build systems related to tests.

  1. Discussion about Github webhooks and Github API

The webhooks and the github API were also mentioned in the meeting given that in order for the builds to be triggered there must be a webhook setup to either build each PR/commit automatically or with a command posted by an authorized user or allow both.

We also talked about the joomla github API which provides methods to communicate with the github API and which can be used in the project  

  1. Discussion about swarm and first scripts

Regarding swarm and the first scripts I did the first tests initializing a swarm with manager nodes and trying to build the scripts to initialize the swarm, some problems were encountered with those scripts mainly because with the exec() or shell_exec() methods some errors came up because the shell commands were ran separately and not in the same exec() command (not in the same shell “lifespan”). Robo was mentioned in order to try and solve those problems with the shell and implement those swarm initialization scripts as robo commands/tasks. The work with the first scripts to initialize the docker swarm was pushed into github to my fork to a specific branch (https://github.com/rjcf18/gsoc17_pr_testing_platform/tree/swarm-init).