flipt-io / flipt Goto Github PK

Increase aggregate test coverage to > 80%

1. Update constraint or variant (any property)
2. Click Save
3. Notice it doesnt reflect in UI
4. Refresh
5. Update is not visible

What

I'd like to be able to export all flags, variants, segments, constraints and rules to a file (YAML/TOML/your favorite ML).

This would allow storing all configuration in source control as well as provide a simple way to backup data.

I'd also like to be able to import previously exported data

I'm thinking this would require two new commands flipt import and flipt export with options for the file name and type (yaml/toml/json/etc).

Open Questions:

1. What to do when there is conflicting data/data already exists during import?

It could make sense to add an option to drop all existing data before importing by providing a -drop or -delete option such as https://github.com/Unleash/unleash/blob/master/docs/import-export.md

Case: I have 10 000 (and increasing) variants for one flag. In my code i wan't manualy write code for each variant key. I want get variant contain json with variant data. Example variant json:

{
"color": "black",
"discount": 3,
"category_id": 24
}

And in my code write one code when handle json from variant instead if/else/switch for each variant.

Describe the solution you'd like
Add pagination to all table components in UI (ex: defaulting to 15 elements per page)

Additional context
https://buefy.org/documentation/table/#pagination-and-sorting

Use: https://buefy.org/documentation/table/#async-data

Is your feature request related to a problem? Please describe.
Add some end to end UI tests to verify that changes do not break the Flipt UI and that users see the expected results.

Describe the solution you'd like
Perhaps something like testcafe? I dont know much about the 'modern' e2e Javascript testing frameworks.. but I know things I DO NOT WANT:

❌ No ❌ :

• Selenium or browser based tests as these are slow and error prone
• BDD testing language, just want simple easy to write expectation based tests

Additional context
Something that could capture screenshots when tests fail would be nice too for local development

Add caching layer on the server side to increase evaluation speed.

Should probably be an LRU cache, perhaps something like Groupcache LRU

Should be able to be disabled and maybe swappable for other implementations in the future

Any plans on expanding this to use external databases, like mysql or postgres? It's the coolest feature flag system I've seen. :)

Is your feature request related to a problem? Please describe.
I need to toggle a feature in response to a certain parameter, e.g. I have 5 admin emails and if the API call sends one of them in the validate call, the feature should be toggled on by responding with true.
Currently I create a segment for each with the constraint email to be equal and the flag returns true for these segments. But setting several emails to the segment as constraints, since it checks if the given email is equal to all of them does not work. Only if just one email is provided.

Describe the solution you'd like
Maybe set the option of the segment from always match all to be switchable to match one of. That way, I would only need to create one segment, give them all the emails as constraints and if the email matches one of the constraints, its in that segment and the feature flag would return true.

Additional context
I really enjoy your work and it is a great help. Thank you.

I add 280 variants to one flag.
I go to Targeting tab and click New Rule.
I select A Percentage Rollout.
Interface draw my 280 variants with fre-filled 0,36%
I click Add Rule and get error Total percentage cannot exceed 100%.

Because 280 * 0,36 = 100,8

Flipt incorrectly calculated the initial values.

Dependabot couldn't find a Gopkg.toml for this project.

Dependabot requires a Gopkg.toml to evaluate your project's current Go dependencies. It had expected to find one at the path: /Gopkg.toml.

If this isn't a Go project, or if it is a library, you may wish to disable updates for it from within Dependabot.

You can mention @dependabot in the comments below to contact the Dependabot team.

I think i've stumble upon another issue...so here's what I did:

• Created a rule test;
• Created a segment test with a constraint: @alwaysMatch IS NOT PRESENT
• Added two variants to test (on, off);
• Modified the targeting of test to be 50% off, 50% on;
• Deleted the two variants individually, created 5 new ones (a, b, c, d, e)—deleted the previous targeting and added a new one, whereas a gets 50% and b, c, d, e gets 10% respectively.

Issue:

{
    "requestId": "1e5aae90-4ea3-4311-be85-91d0d24ca3e4",
    "entityId": "3",
    "requestContext": {},
    "match": true,
    "flagKey": "test",
    "segmentKey": "test",
    "timestamp": "2019-05-11T18:22:53.517088460Z",
    "requestDurationMillis": 0.980012
}

I've tested programatically with 100 different IDs, for some reason, it doesn't evaluate to anything with some of them.

State of the world:

sqlite> .tables
constraints        flags              schema_migrations  variants
distributions      rules              segments

sqlite> select * from flags;
test|test||1|2019-05-11 18:09:34.824808692+00:00|2019-05-11 18:09:34.824808692+00:00

sqlite> select * from rules;
09a3cee0-448a-4100-8d55-b2a68df41eae|test|test|1|2019-05-11 18:22:02.831441605+00:00|2019-05-11 18:22:02.831441605+00:00

sqlite> select * from segments;
test|test||2019-05-11 18:09:39.674672314+00:00|2019-05-11 18:09:39.674672314+00:00

sqlite> select * from distributions;
fe817dd0-fc27-4d30-b0d5-ccd4ea0a3282|09a3cee0-448a-4100-8d55-b2a68df41eae|af3c2070-9692-40b7-9258-608c24a53b8d|50.0|2019-05-11 18:22:02.842112509+00:00|2019-05-11 18:22:02.842112509+00:00
ee4743af-c6fc-4523-b38a-03c1713b7d1b|09a3cee0-448a-4100-8d55-b2a68df41eae|d7049aa6-9906-4e9d-b6e5-9a9be424992a|10.0|2019-05-11 18:22:02.846632236+00:00|2019-05-11 18:22:02.846632236+00:00
c43b7cb4-0c19-4e12-811d-2f4c774cc506|09a3cee0-448a-4100-8d55-b2a68df41eae|e019aff0-df11-40da-878d-5b4293c69f15|10.0|2019-05-11 18:22:02.846011664+00:00|2019-05-11 18:22:02.846011664+00:00
0c4e8daf-9f34-4691-9e6f-23c3a801ff10|09a3cee0-448a-4100-8d55-b2a68df41eae|8fce46af-045b-4b34-a4bb-c362585908f4|10.0|2019-05-11 18:22:02.842992967+00:00|2019-05-11 18:22:02.842992967+00:00
e4072afe-0959-4ab0-970c-106f6ef9754a|09a3cee0-448a-4100-8d55-b2a68df41eae|59e512a4-f486-4b4c-9857-28e4e759fa1d|10.0|2019-05-11 18:22:02.844072062+00:00|2019-05-11 18:22:02.844072062+00:00

sqlite> select * from variants;
af3c2070-9692-40b7-9258-608c24a53b8d|test|a|a||2019-05-11 18:21:28.402943187+00:00|2019-05-11 18:21:28.402943187+00:00
8fce46af-045b-4b34-a4bb-c362585908f4|test|b|b||2019-05-11 18:21:32.100555715+00:00|2019-05-11 18:21:32.100555715+00:00
59e512a4-f486-4b4c-9857-28e4e759fa1d|test|c|c||2019-05-11 18:21:34.568535072+00:00|2019-05-11 18:21:34.568535072+00:00
e019aff0-df11-40da-878d-5b4293c69f15|test|d|d||2019-05-11 18:21:37.8885922+00:00|2019-05-11 18:21:37.8885922+00:00
d7049aa6-9906-4e9d-b6e5-9a9be424992a|test|e|e||2019-05-11 18:21:41.021846003+00:00|2019-05-11 18:21:41.021846003+00:00

Hi! Thanks for the flipt!

I have a need to give specific variant in a limited quantity, for example, 1000.

I would also like to do 2 distribution strategies:

1. start distributing the variant to everyone with the highest priority until quantity ends.

2. evenly adding for a certain time, but not more than a given quantity.

MySQL is still a popular DB, so let's support it

Would like a better logo and favicon but I'm not good at design ☹️

Something that represents feature flagging? flags? toggles?

Describe the bug

User can input any flag key/segment into the url, even those that don't exist, and will be presented with a 'normal looking' view.

We should handle the 404 being returned by the API and show a warning/error message, and probably redirect to the flags index or segments index respectively.

Version Info
Run flipt --version and paste the output here.

Version: v0.10.1-snapshot
Commit: 7be211afb564df98bcf6a54fb0511733c8f00fe2
Build Date: 2019-11-10T15:41:33Z
Go Version: go1.13.3

To Reproduce

Type http://{flipt_server_url}/#/flags/doesntexist into browser and hit enter

Expected behavior

• Error or warning message pops up about flag/segment not found
• User is redirected to flag or segment index page

Screenshots

Logs

time="2019-11-10T15:42:11Z" level=info msg="finished unary call with code NotFound" error="rpc error: code = NotFound desc = flag \"bar\" not found" grpc.code=NotFound grpc.method=GetFlag grpc.service=flipt.Flipt grpc.start_time="2019-11-10T15:42:11Z" grpc.time_ms=2.185 peer.address="127.0.0.1:51684" server=grpc span.kind=server system=grpc

Describe the solution you'd like

Would like to support evaluation caching for rule based evaluation. IE https://flipt.dev/concepts/#evaluation

This would behave similarly to what was added in #6, but would likely be more complex since there are multiple entities that need to be cached (segments, distributions, rules, variants, etc) and invalidated.

One approach is to just cache the evaluation result and not the actual individual actors (noted above), but would have to come up with a scheme to invalidate this cached result if any of the above changed.

https://github.com/getsentry/flipt-grpc-python

just an fyi in case you wanna document it. can add you to the repo.

also added a Makefile which may be useful in other repos?

we're exploring using flipt so this is needed for our eval

Describe the bug
Adding a flag with a : character breaks the UI. e.g. foo:bar

Problem

When creating a new flag/segment/etc, you have to type the name and key field before saving as they are both required.

Purposed Solution

It would be nice if the key field autopopulated based on the name fields as they are often closely related.

Example

^ In the above screenshot, it would be nice if key was autopopulated to foo-bar

Also consider if it makes sense to swap the position of the name and key fields in the form

Is your feature request related to a problem? Please describe.

I'm currently using unleash, but I'm considering flipt. I'm going through the docs and noticed that migrations are kept in the db.migrations.path config, which points to a local file. There is no indication that migrations are stored locally, but I'd love to be wrong.

Edit: Are these files similar to migration files in rails or are they used to keep track of what migrations have been run? If the former, then I assume the DB does keep track of migrations.

Describe the solution you'd like

A way to store all data in a db, so that I don't need to configure permanent storage for my ECS/Fargate setup.

Additional context
N/A

It would be great if we could handle unmarshalling errors more gracefully during evaluation.

Example

Ideas

1. Catch unmarshalling errors and return a nicer error that informs the user that values must be strings
2. Infer "true" from true and 'autobox' the type into a string

Hey @markphelps—great work.

I'm thinking about using flipt for a product I'm developing, however, I am not sure if I'm doing something wrong.

I've created one flag called "test" with a segment that matches on all contexts and a targeting rule on set to 50% and off also set to 50%. I can't seem to get a match: true or match: false back from the server.

The next release of go will probably enable modules by default. We should update Flipt to remain current and use modules.

I previously tried to do this and ran into some issues with GRPC or GRPC Gateway (I can't remember), so I would like to try again and document any issues if there are still any blockers.

It seems grpc-go should support modules just fine: https://github.com/grpc/grpc-go#installation

Going to try the approach here: https://blog.callr.tech/migrating-from-dep-to-go-1.11-modules/

Is your feature request related to a problem? Please describe.
Feature, allow evaluate all available flags

Describe the solution you'd like
Querying available flags which is return all flags that fit for current context. Example:

GET /api/v1/evaluate_all
{
  "entity_id": "string",
  "context": {
    "property1": "string",
    "property2": "string"
  }
}

and returns

{
  "entity_id": "string",
  "request_context": {
    "property1": "string",
    "property2": "string"
  },
  "match_flags": [
      {
        "key": "string",
       "value": "string"
      }
  ]
}

Additional context

Add a Heroku button to allow for someone to easily standup an example app in Heroku:

https://devcenter.heroku.com/articles/heroku-button

I have multiple databases (aka tenant) since I developed a multi-tenancy system, every tenant is a customer who owns a site https://[tenant_name].mysystem.com, how can I smoothly toggle features for a specific tenant using Flipt?

Re: #41

Add support for external DB storage with Postgres.

Requirements:

• Satisfy the same interface in the storage package.
• Be configurable, meaning the user can swap out the default SQLite implementation with Postgres via the config file. This will likely mean changing the config format for db, ex:
  https://github.com/markphelps/flipt/blob/3d4f3f02f7634d88d0219a2f733f5298dd0fc7d5/config/default.yml#L18
  As path doesn't make sense for a Postgres database, I'm thinking something like this for Postgres:

db:
  url: postgresql://localhost:5432/flipt

And changing the default SQLite config to look something like this:

db:
  url: sqlite3:///var/opt/flipt/flipt.db

Or something similar, still on the fence about what would work best here. Feedback would be greatly appreciated!

• Use github.com/Masterminds/squirrel
• Handle foreign key/constraint errors like: https://github.com/markphelps/flipt/blob/3d4f3f02f7634d88d0219a2f733f5298dd0fc7d5/storage/flag.go#L234

This may make sense to abstract away the error handling since that is what is most likely to be different between the different databases. We might be able to get away with not changing the actual query logic much at all since squirrel and Go's sql lib should do most of the heavy lifting.

Maybe not a top priority right now...but do you think there's a way to batch-evaluate a set of flags?

I plan to use the REST API instead of grpc—so like, it'd be nice if I could send a single request with a set of flags to be evaluated instead of 1 request per flag. Just to avoid all the roundtrips.

Thoughts?

Describe the bug
When creating a segment with two constraints:

• boolean constraint == true
• and a string == 'somevalue'

The segment is matched when evaluating a flag with the boolean value only.

Version Info
Run flipt --version and paste the output here.

Version: 0.8.0
Commit: 0a47977b560085983cd95eb5f31141039df8ecd2
Build Date: 2019-09-30T08:43:15Z
Go Version: go1.13

To Reproduce

1. Create a flag with one variant
2. Create a segment
   2a. Create a constraint isAdmin, boolean and == TRUE
   2b. Create a constraint customer, string and equals acme
3. Evaluate the created flag with the following contexts:

{
    "flagKey": "flag",
    "entityId": "ca93e3ad-8327-4da6-909c-f5f7b2c9a4f5",
    "context": {
        "isAdmin": "true"
    }
}

Expecting no match

{
    "flagKey": "flag",
    "entityId": "ca93e3ad-8327-4da6-909c-f5f7b2c9a4f5",
    "context": {
        "isAdmin": "true",
        "customer": "acme"
    }
}

Expecting a match

{
    "flagKey": "flag",
    "entityId": "ca93e3ad-8327-4da6-909c-f5f7b2c9a4f5",
    "context": {
        "isAdmin": "true",
        "customer": "apple"
    }
}

Expecting no match

Additional context

• Running the latest docker container
• Database used Postgres

I could use an easy way to turn feature flags on and off based on segments.

As an example, my web application has a feature called "projects" that I would like to enable only for users that have subscribed to the "professional" plan. As far as I know, the only way to do that in flipt is to create two variants of the flag ("enabled" and "disabled"), and to map those variants to segments corresponding to the plans ("basic" or "professional").

Having to create those variants is a bit inconvenient as all I really want to do is just switch the flag on or off.

Is your feature request related to a problem? Please describe.
Since my systems already use Etcd as a storage backend, an Etcd backend for Flipt would make it easier to deploy in my environment.

Describe the solution you'd like
I would like an Etcd storage backend, in addition to the PostgreSQL and SQLite options.

Additional context
Adding an Etcd backend will be a shift from the SQL backends currently offered, but hopefully any work towards Etcd would also be useful towards supporting other KV backends, like Redis.

Describe the bug
When requesting a feature flag, with the same request entityId and payload and given two rules:

1. Percentage rollout over a specific segment -> 50% released 50% unreleased
2. Match all -> unreleased

We get inconsistent match and results over sequential requests.

Version Info
0.10.0

To Reproduce
On a fresh install

1. Create and enable a new feature flag
2. Create two variants released and unreleased
3. Create a segment subscribers that checks "premium_user": "true"
4. Create a segment all users that match all
5. In the previously created feature flag, setup two rules (in this order):
   1. If Matches segment subscribers then serve percentage rollout -> 50% released 50% unreleased
   2. If Matches segment all users then serve unreleased

Expected behavior
This is a legitimate use case where we want to progressively roll-out a feature to premium users while keeping it disabled for all others.

The expected behaviour when requesting this feature flag as a subscriber (i.e. "premium_user": "true") with the same payload is to have a consistent match on the first rule.

But currently, with the exact same payload (i.e. same entityId and context) we get randomly matched on first or second rule, effectively returning an inconsistent variant released or unreleased for one given premium user.

Screenshots

Request A -> Match Percentage Roll out (rule 1)

Request B -> Match all (rule 2)

Please note the inconsistent value of fields segmentKey and value in the response.

Additional context

• Database used: Postgres

For backup the sqlite database I can use flipt export which pipes to STDOUT. In our k8s environment this looks so:
kubectl exec -it <name of flipt pod> -- /bin/sh -c './flipt export' > export.yaml
... and works like a charm :-)

But now I wonder how to import the the backup since flipt import wants to read from a file, not from STDIN. I've tried a lot of variants, all ended in Error: requires a file argument :-(

So it would be great, if I can do a flipt import to read from STDIN for restoring backups in a docker or k8s environment.

Describe the bug

Using the wonderful auto-population of key fields for flags/segments that was added in #155, I noticed that any extra spaces before/after/between words in the name field correspond to extra -'s in the key field.

It would be great if we could strip these extra -s from the key field.

This will allow us to add stricter validation around the key field on the backend

Version Info
0.10.0

To Reproduce
See gif below ⬇️

Expected behavior

• Extra whitespace in the name field should be trimmed before/after the name which means the key field will not contain extra -s
  • Ex: name: foo should become name: foo, key: --foo-- will become key: foo
• Extra spaces in the name field between words should ideally only result in a single - in the key:
  • Ex: name: foo bar should result in key: foo-bar, not foo---bar

Screenshots

/cc @aaronraff

IMO there isn't great documentation on what you can do with variants, segments, rules and distributions.

It might make sense to add more examples that show how this could be used, much like in examples/basic, but examples/targeting?

Also would be nice if these abilities could somehow be expressed better in diagrams or even just in words, kind of like what LaunchDarkly does with targeting users, but maybe not as verbose.

Initial settings:

I request evaluate:

{
"flag_key": "demo",
"entity_id": "1"
}

and gets:

{
  "requestId": "fe7c640a-8715-4a08-87d3-cfdb8d014f6d",
  "entityId": "1",
  "match": true,
  "flagKey": "demo",
  "segmentKey": "non-registered-visitors",
  "timestamp": "2019-11-24T10:33:43.123944900Z",
  "value": "three",
  "requestDurationMillis": 0.262
}

Then I modified percentage rollout (set first = 0):

And again request evaluate:

{
"flag_key": "demo",
"entity_id": "1"
}

and gets:

{
  "requestId": "dc1939ba-081d-40f5-afbe-d8a3c1136d84",
  "entityId": "1",
  "flagKey": "demo",
  "segmentKey": "non-registered-visitors",
  "timestamp": "2019-11-24T10:34:27.103771200Z",
  "requestDurationMillis": 0.293
}

Why did the value disappear?

Is your feature request related to a problem? Please describe.
It would be nice to provide an export command that allows users to export the current state of their database in a yaml or json format.

This would be helpful when debugging issues as I could ask users to attach the contents or subset of their flipt export data to bug reports.

Describe the solution you'd like
Support flipt export [--format=json | --format=yaml] to export data to a local file

Additional context
Add any other context or screenshots about the feature request here.

Evaluation gives runtime error if distribution doesnt add up to 100%:

"error": "rpc error: code = Internal desc = runtime error: index out of range"

Is your feature request related to a problem? Please describe.
Currently logs only go to STDOUT, would be great to support logging to the filesystem

Describe the solution you'd like
Support logging to file in configs

It's not obvious to a user what "on" or "off" does. I would assume they fully turn the feature on or off, but at least the for the on state thats not true.

So here's the question:

Should a flag with "on" and no rules be on? If not, then maybe its at least worth clarifying in the user (help text? something else?) what the behavior is.

Describe the bug
The variants column in the flags list always says yes, though my assumption is it should only say yes if there are variants attached to a flag.

Version Info
0.13.0

Version 0.6.0
Binary installed to ~/flipt

Default configuration with exception of the sqlite3 db file located in ~/flipt, and migrations pointing to ~/flipt/config/migrations/sqlite3

Running ./flipt results in error that ~/flipt/config/migrations/sqlite3 file does not exist.

Copying migrations from this repo into the migrations folder resolves the issue.

The last build to include migrations was 0.4.2.

Some dev dependencies are several versions behind. Would like to get them as up to date as possible.

~/workspace/flipt/ui master
❯ yarn outdated
yarn outdated v1.17.3
info Color legend :
 "<red>"    : Major Update backward-incompatible updates
 "<yellow>" : Minor Update backward-compatible features
 "<green>"  : Patch Update backward-compatible bug fixes
Package                            Current Wanted  Latest Package Type    URL
babel-loader                       7.1.5   7.1.5   8.0.6  devDependencies https://github.com/babel/babel-loader
copy-webpack-plugin                4.6.0   4.6.0   5.0.3  devDependencies https://github.com/webpack-contrib/copy-webpack-plugin
css-loader                         0.28.11 0.28.11 3.0.0  devDependencies https://github.com/webpack-contrib/css-loader
eslint                             5.16.0  5.16.0  6.0.1  devDependencies https://eslint.org
eslint-config-prettier             3.6.0   3.6.0   6.0.0  devDependencies https://github.com/prettier/eslint-config-prettier#readme
eslint-config-standard             10.2.1  10.2.1  13.0.1 devDependencies https://github.com/standard/eslint-config-standard
eslint-friendly-formatter          3.0.0   3.0.0   4.0.1  devDependencies https://github.com/royriojas/eslint-friendly-formatter#readme
eslint-loader                      1.9.0   1.9.0   2.2.1  devDependencies https://github.com/webpack-contrib/eslint-loader#readme
eslint-plugin-node                 5.2.1   5.2.1   9.1.0  devDependencies https://github.com/mysticatea/eslint-plugin-node#readme
eslint-plugin-promise              3.8.0   3.8.0   4.2.1  devDependencies https://github.com/xjamundx/eslint-plugin-promise#readme
eslint-plugin-standard             3.1.0   3.1.0   4.0.0  devDependencies https://github.com/standard/eslint-plugin-standard#readme
file-loader                        1.1.11  1.1.11  4.0.0  devDependencies https://github.com/webpack-contrib/file-loader
html-webpack-plugin                2.30.1  2.30.1  3.2.0  devDependencies https://github.com/jantimon/html-webpack-plugin
optimize-css-assets-webpack-plugin 3.2.1   3.2.1   5.0.3  devDependencies http://github.com/NMFR/optimize-css-assets-webpack-plugin
ora                                1.4.0   1.4.0   3.4.0  devDependencies https://github.com/sindresorhus/ora#readme
portfinder                         1.0.20  1.0.21  1.0.21 devDependencies https://github.com/indexzero/node-portfinder#readme
postcss-import                     11.1.0  11.1.0  12.0.1 devDependencies https://github.com/postcss/postcss-import#readme
postcss-loader                     2.1.6   2.1.6   3.0.0  devDependencies https://github.com/postcss/postcss-loader#readme
postcss-url                        7.3.2   7.3.2   8.0.0  devDependencies https://github.com/postcss/postcss-url#readme
semver                             5.7.0   5.7.0   6.2.0  devDependencies https://github.com/npm/node-semver#readme
shelljs                            0.7.8   0.7.8   0.8.3  devDependencies http://github.com/shelljs/shelljs
uglifyjs-webpack-plugin            1.3.0   1.3.0   2.1.3  devDependencies https://github.com/webpack-contrib/uglifyjs-webpack-plugin
url-loader                         0.5.9   0.5.9   2.0.1  devDependencies https://github.com/webpack-contrib/url-loader
vue-loader                         13.7.3  13.7.3  15.7.0 devDependencies https://github.com/vuejs/vue-loader
vue-style-loader                   3.1.2   3.1.2   4.1.2  devDependencies https://github.com/vuejs/vue-style-loader#readme
webpack                            3.12.0  3.12.0  4.36.1 devDependencies https://github.com/webpack/webpack
webpack-dev-server                 2.11.5  2.11.5  3.7.2  devDependencies https://github.com/webpack/webpack-dev-server#readme
   Done in 1.07s.

Figure out a way to test Flipt from outside.

Possible options:

• Run flipt in a separate process and test using the GRPC and/or REST APIs
• Run flipt in a separate process and test via the UI using headless chrome?
• ?

I think this query should be ordered by created_at evaluations same as rules

Add support for users wanting to use their own HTTPS certs

Great project. I'd like to contribute, do you have features that you have in mind? If so, it would be great if you can create issues so contributors can jump in.

When trying to create a variant for a flag where the key is default, I'm getting an error suggesting that the key needs to be unique, despite the no existing variants for the flag. I do have a variant with the key default on another flag though.

I would expect that the keys for variants should be unique per flag rather than globally, since there is no way that these variants can be used across flags.