Mongoengine: Travis tests for python 2.6 are failing

[RStatkevych]

Travis show tests failing on python2.6 when I was trying to submit my pull request [#1282] . First I thought that this is my fault, but I see that that few recent builds have the same fails as mine (for example https://travis-ci.org/MongoEngine/mongoengine/jobs/128043817 and mine is https://travis-ci.org/MongoEngine/mongoengine/jobs/129869183). What the problem can it be? Is this travis problem or mongoengine?

P.S: I noticed that it's not reported or asked yet, so let it be here

[wojcikstefan]

Yup, noticed that, too. Few thoughts about this:

• It appears that the error comes from the rednose package. I looked at https://github.com/JBKahn/rednose to see if it supports Py v2.6 , but don't see anything.
• This is an example of adding a dependency that's not really necessary - it comes at a price, and IMHO the price is higher than the value it provides (some nice colors in the nosetests' output).
• We we even want to keep Py2.6 support? Personally I'd push for dropping it sooner rather than later.

@lafrech @thedrow what do you think?

[thedrow]

We really should switch to py.test.
It's more maintained and has better plugins.
As far as I'm concerned we can drop 2.6 support.

[lafrech]

I've been dealing with code tested with pytest recently and I can't tell the difference. I don't know what it would take to make the transition. In fact, in my recent projects, I've been using nosetests... because I copid MongoEngine... No objection to this change if everyone feels like porting the tests. Should we file a dedicated issue ?

Regarding Python 2.6, I don't mind. The least old versions we support, the easiest. I guess the rationale is that some Linux distros like Red Hat will still ship 2.6 for about 5 years. But users can install 2.7 and I have no idea who actually uses 2.6 and who would be totally blocked (can't move to 2.7 for any platform or other lib issue). It's quite possible that users already moved to 2.7 for another dependency.

Considering current effort put into MongoEngine and bugs piling up almost daily, 2.6 support might be a luxury so it could be justified to drop it. If anyone really needs it and is willing to work on it, it's never too late to move back. Meanwhile, we may remove it from the tests and let it go.

As to the root cause of the rednose issue, I don't know if it breaks while processing working code, or while dealing with an error in MongoEngine.

[thedrow]

py.test should be able to run nose tests as well.

[wojcikstefan]

So we have +3 on dropping python 2.6 support (which, at least for now, would only include dropping it from the tox config) - is there anybody else that should give a thumbs up before this decision is made?

As to the root cause of the rednose issue, I don't know if it breaks while processing working code, or while dealing with an error in MongoEngine.

Yea, I don't know as well, but it's bad either way... It either covers up the actual error or it makes a successful test fail. Anyway, it doesn't seem to be an issue in Py 2.7+.

[lafrech]

Let's ping @MongoEngine/mongoengine team, just in case...

[lafrech]

Closed by https://github.com/MongoEngine/mongoengine/pull/1391.