Bump. I ran into more cx-freeze issues again with 1.5.1. Every new version of MPL causes me hours of debugging to get cx-freeze working again! ARGH! In this case I had to revert to 1.3.1.

How can we get this working?

PRs to add tests are always welcome. On the master branch we now run tests on appveyor so we will get both linux and windows coverage on the test.

I am not sure if this sort of thing fits naturally into the unit test framework, if you have a minimal script that uses cx_freeze with mpl, can you put it in the ci top level folder and run it from the travis and appveyor configs?

I spent a few hours trying to get a simple script to package in cx_freeze
last night and ultimately gave up, finding that pyinstaller worked for my
project without any messing about.

It would still be good if we could add a test. I can put together an
example, but I have no idea how to force it to fire from the Travis or
Appveyor builds

On Thu, Mar 17, 2016 at 9:32 AM, Thomas A Caswell notifications@github.com
wrote:

PRs to add tests are always welcome. On the master branch we now run tests
on appveyor so we will get both linux and windows coverage on the test.

I am not sure if this sort of thing fits naturally into the unit test
framework, if you have a minimal script that uses cx_freeze with mpl, can
you put it in the ci top level folder and run it from the travis and
appveyor configs?

—
You are receiving this because you commented.
Reply to this email directly or view it on GitHub
#3391 (comment)

I'm nervous about the idea of putting a test for cx_freeze in Travis. If a changeset makes the test fail, how many people are going to have the slightest idea of what to do about it--or whether it is a false alarm, given that we already have quite a bit of trouble with those?

I certainly see your point. But it is crucial that matplotlib be able to package into an exe in some fashion or another. Ideally in both cx-freeze and PyInstaller. I frequently see problems with both packaging tools with MPL, though PyInstaller does seem the more mature option currently.

Maybe make it an "allow fail" test, like we do for the nightly python
builds?

On Fri, Mar 18, 2016 at 10:39 AM, Ian Bell notifications@github.com wrote:

I certainly see your point. But it is crucial that matplotlib be able to
package into an exe in some fashion or another. Ideally in both cx-freeze
and PyInstaller. I frequently see problems with both packaging tools with
MPL, though PyInstaller does seem the more mature option currently.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
#3391 (comment)

If we want it to stay working, it needs a test. If it breaks a bunch, we need to figure out how to make it not.

Echoing @WeatherGod I suggest adding it to an allowed failure test in the beginning. If it works fine we can move it to a regular test.

I have no idea what requirements must be met for freezing to work. Would someone provide a link, please? And maybe an example of a change that broke freezing, and how it was fixed? If a test is added, then I suggest that this information must be added to the Developer section of the mpl docs at the same time. Or better, it could be added right away.

Is this something that is specific to travis (in that case we can close) or should it be renamed to CI (or something)?

This issue has been marked "inactive" because it has been 365 days since the last comment. If this issue is still present in recent Matplotlib releases, or the feature request is still wanted, please leave a comment and this label will be removed. If there are no updates in another 30 days, this issue will be automatically closed, but you are free to re-open or create a new issue if needed. We value issue reports, and this procedure is meant to help us resurface and prioritize issues that have not been addressed yet, not make them disappear. Thanks for your help!

I am going to close this as won't fix.

I suspect we would merge a weekly cron job to to this test, but probably do not want to run it on every PR.