Skip to content

Conversation

@krisr
Copy link
Contributor

@krisr krisr commented Jul 3, 2013

This change disables new uses of the percent and percent_visited strategies and also introduces the new aliases percent_deprecated and percent_visited_deprecated.

The goal is to switch all current uses of the percent based strategies over to the *deprecated alias names after this change goes out. This will enable us to follow up with another commit that corrects the implementation of the percent based strategies without affecting the set of users that are exposed to the deprecated implementation of the strategies.

We hope to re-implement the percent based strategies in a way that can be used safely with experiment analysis. The current implemention tends to distribute features using the percent based strategies over the same subsets of users making it difficult to draw conclusions from the results of the analysis of those features.

Note that my editor strips trailing whitespace. Not sure what people think about that yet ;).

@mcharkov it would be great to get you to review this.

Testing:
I wrote new specs and verified that all specs passed

…tegy

This change disables new uses of the percent and percent_visited strategies and also introduces the new aliases percent_drecated and percent_visited_deprecated.

The goal is to switch all current uses of the percent based strategies over to the *deprecated alias names after this change goes out. This will enable us to follow up with another commit that corrects the implementation of the percent based strategies without affecting the set of users that are exposed to the deprecated implementation of the strategies.

We hope to re-implement the percent based strategies in a way that can be used safely with experiment analysis. The current implemention tends to distribute features using the percent based strategies over the same subsets of users making it difficult to draw conclusions from the results of the analysis of those features.

Testing:
I wrote new specs and verified that all specs passed
Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Might be clearer to break this out into two tests that each have one assertion (even if they both just call should_only_launch_to_a_percentage_of_users).

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I thought about that, but seems overkill for a temporary commit that is just being used until I ship the next version that fixes the percent implementation.

@lkosak
Copy link

lkosak commented Jul 3, 2013

LGTM, especially if this stuff is just temporary.

Copy link

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this also temporary? Our ruby style guide sez
"Avoid the usage of class (@@) variables due to their "nasty" behavior in inheritance."

@jasonkb
Copy link

jasonkb commented Jul 3, 2013

LGTM

Delightful to see trebuchet get love!

@mcharkov
Copy link
Contributor

mcharkov commented Jul 3, 2013

Looks good to me!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants