Botfront.
 Github
Getting Started
Configuring channels
Authoring conversations
Conversational interactions
Natural language understanding
Server installation
Developers
Administration
Misc
Migration guideMinor versionsMajor versions

Migration guide

As we haven’t reached 1.x we consider minor versions as major and they could have breaking changes.

Minor versions

For example 0.18.1 -> 0.18.3 Botfront will offer you to upgrade your project. Part of the process is automated, buty you will still have ONE thing to do manually: Take:

  • Take rasa-for-botfront image version in .botfront/botfront.yaml. In the following it would be v1.5.1-bf.10 (with the v)
  • Take rasa-sdk image version in .botfront/botfront.yaml. In the following it would be 1.3.2 (without the v)
images:
default:
...
rasa: botfront/rasa-for-botfront:v1.5.1-bf.10
actions: rasa/rasa-sdk:1.3.2
...

And apply these versions to:

  • rasa/Dockerfile
  • actions/Dockerfile
  • actions/Dockerfile.production

Major versions

0.18.x -> 0.19.x

Breaking changes

We significantly changed (and improved) how bot responses work. Only three types remain:

  • Text
  • Text with buttons
  • Image
  • Custom

If you were using other response types (mostly facebook related such as Templates, Lists, …) they will be lost. You need to make a copy before upgrading.

Upgrading a project

The project structure has changed

You will need to create another project and copy your data in it:

  1. Create a new project with botfront init.
  2. Copy the botfront-db folder from your old project to the newly created project. Make sure to copy and not move your db so you can always recover it from your existing project. Your existing project should remain unchanged.
  3. If you have custom actions, copy them to the actions folder in the new project.
  4. Run your project with botfront up. At this point, you should be able to log in.
  5. Go to Settings > Endpoints If you are running with the default CLI configuration, replace:
nlg:
url: "http://botfront-api:8080/project/bf/nlg"

with:

nlg:
type: "rasa_addons.core.nlg.GraphQLNaturalLanguageGenerator"
url: "http://botfront:3000/graphql" # This should be the same host as the Botfront app
  1. Restart Rasa with botfront restart rasa
  2. Train
  3. You’re done, have fun :)
🖊️ Edit this page on Github