Project

General

Profile

Actions

action #1952

closed

Get travis.ci and coveralls running and working

Added by _miska_ about 10 years ago. Updated about 10 years ago.

Status:
Resolved
Priority:
Urgent
Assignee:
Category:
-
Target version:
Start date:
2014-03-20
Due date:
% Done:

100%

Estimated time:
40.00 h

Description

There is some initial work done regarding travis.ci. We have installation of perl modules in place and we have a module that can run tests and report them to coveralls. What currently doesn't work is running the tests. My findings so far, thanks goes to Ancor for pointing out where the problem might be.

Problem is following function in lib/OpenQA.pm
101: has secrets => sub {

The function that is called there to initialize secrets is called before schema is initialized. I tried using global variable and wait in this function for it to get true (which only happened in has schema => sub) and that didn't happened.

I tried assigning to it something meaningful only after we have connection to database set, but somehow it didn't worked (either I got syntax error or everything was still using empty array)

I tried making this method lazy, but I got from tests lazy as a string when they failed, so I guess this didn't worked out somehow as I wanted.

I also tried to convert it into function that would return empty array initially and correct results once database is set up, didn't worked either (same tests as previously were broken).

Actually if I simple break the credentials, I get quite some tests passing, not all of them, but quite some.

Actions #1

Updated by _miska_ about 10 years ago

Oh, and one more thing, probably most important thing, Ancors theory was that

507 CORE::state $schema;

in lib/OpenQA/modules/openqa.pm doesn't cope well with forking

Actions #2

Updated by aplanas about 10 years ago

  • Status changed from New to In Progress
Actions #3

Updated by aplanas about 10 years ago

  • % Done changed from 0 to 50

There where several problems

  • Mojo 4.90 change the moment that recover the 'secret' attribute. This cause that there is an access of the database before we have a in memory database. This issue is published by lnussel:

https://github.com/kraih/mojo/issues/607

  • To avoid the creation of the Mojo applications, we use in Travis 'prove' instead of 'make test'

  • One test fails because internally create the mojo application creating a Test:Mono instance before creating the :memory: database

Actions #4

Updated by aplanas about 10 years ago

  • Status changed from In Progress to Resolved
  • % Done changed from 50 to 100
Actions

Also available in: Atom PDF