Search: Fix whitespaces

[3v0k4]

Fixes: basecamp/kamal#1176

---

Using strip_newlines formatted content into (notice the missing whitespaces):

"Kamal 2: Configuration changesBuilderThe builder configuration has been simplified.ArchYou must specify the architecture(s) you are building for:# single archbuilder:  arch: amd64# multi archbuilder:  arch:    - amd64    - arm64Remote buildersSet the remote directly with the remote option. By default, it will only be used if the arch you are building doesn’t match the local machine:builder:  arch: amd64  remote: ssh://docker@docker-builderYou can force Kamal to only use the remote builder, by setting local: false:builder:  arch: amd64  local: false  remote: ssh://docker@docker-builderDriverKamal will now always use the Docker container (link) driver by default. You can set the driver yourself to change this:builder:  driver: dockerThe Docker driver has limited capabilities — it doesn’t support build caching or multiarch images.Traefik → ProxyThe traefik configuration is no longer valid. Instead, you can configure kamal-proxy under proxy.If you were using custom Traefik labels or args, see the proxy configuration to determine whether you can convert them.Be aware that by default kamal-proxy forwards traffic to the container port 80, this is because we assume your container is running Thruster, and it listens on the port 80. If you are running a different service or port, you can configure the app_port setting:proxy:  app_port: 3000kamal-proxy supports common requirements such as buffering, max request/response sizes, and forwarding headers, but it does not encompass the full breadth of everything Traefik can do.If you don’t see something you need, you can raise an issue and we’ll look into it, but we don’t promise to support everything — you might need to run Traefik or another proxy elsewhere in your stack to achieve what you want.HealthchecksThe healthcheck section has been removed.Proxy rolesFor roles running with a proxy, the healthchecks are performed externally by kamal-proxy, not via internal Docker healthchecks. You can configure them under proxy/healthcheck.proxy:  healthcheck:    path: /health    interval: 2    timeout: 2Please note that the healthchecks will use the app_port setting, which defaults to port 80. Previously, healthchecks defaulted to port 3000. You can change this back with:proxy:  app_port: 3000Non-proxy rolesFor roles that do not run the proxy, you can set a custom Docker healthcheck via the options.servers:  web:    ...  jobs:    options:      health-cmd: bin/jobs-healthyFor those containers, Kamal will wait for the healthy status if they have a healthcheck or running if they don’t.You can set a readiness_delay, which is used when we see the running status. We’ll wait that long and confirm the container is still running before continuing.All rolesThere are two timeouts you can set at the root of the config that are used across all roles, whether they use a proxy or not.# how long to wait for new containers to bootdeploy_timeout: 20# how long to wait for requests to complete before stopping old containers# Replaces stop_wait_timedrain_timeout: 20# how long to wait for 'non-proxy role' containers without healthchecks to stay in the running statereadiness_delay: 10"

Replacing it with normalize_whitespace yields:

"Kamal 2: Configuration changes Builder The builder configuration has been simplified. Arch You must specify the architecture(s) you are building for: # single arch builder: arch: amd64 # multi arch builder: arch: - amd64 - arm64 Remote builders Set the remote directly with the remote option. By default, it will only be used if the arch you are building doesn’t match the local machine: builder: arch: amd64 remote: ssh://docker@docker-builder You can force Kamal to only use the remote builder, by setting local: false: builder: arch: amd64 local: false remote: ssh://docker@docker-builder Driver Kamal will now always use the Docker container (link) driver by default. You can set the driver yourself to change this: builder: driver: docker The Docker driver has limited capabilities — it doesn’t support build caching or multiarch images. Traefik → Proxy The traefik configuration is no longer valid. Instead, you can configure kamal-proxy under proxy. If you were using custom Traefik labels or args, see the proxy configuration to determine whether you can convert them. Be aware that by default kamal-proxy forwards traffic to the container port 80, this is because we assume your container is running Thruster, and it listens on the port 80. If you are running a different service or port, you can configure the app_port setting: proxy: app_port: 3000 kamal-proxy supports common requirements such as buffering, max request/response sizes, and forwarding headers, but it does not encompass the full breadth of everything Traefik can do. If you don’t see something you need, you can raise an issue and we’ll look into it, but we don’t promise to support everything — you might need to run Traefik or another proxy elsewhere in your stack to achieve what you want. Healthchecks The healthcheck section has been removed. Proxy roles For roles running with a proxy, the healthchecks are performed externally by kamal-proxy, not via internal Docker healthchecks. You can configure them under proxy/healthcheck. proxy: healthcheck: path: /health interval: 2 timeout: 2 Please note that the healthchecks will use the app_port setting, which defaults to port 80. Previously, healthchecks defaulted to port 3000. You can change this back with: proxy: app_port: 3000 Non-proxy roles For roles that do not run the proxy, you can set a custom Docker healthcheck via the options. servers: web: ... jobs: options: health-cmd: bin/jobs-healthy For those containers, Kamal will wait for the healthy status if they have a healthcheck or running if they don’t. You can set a readiness_delay, which is used when we see the running status. We’ll wait that long and confirm the container is still running before continuing. All roles There are two timeouts you can set at the root of the config that are used across all roles, whether they use a proxy or not. # how long to wait for new containers to boot deploy_timeout: 20 # how long to wait for requests to complete before stopping old containers # Replaces stop_wait_time drain_timeout: 20 # how long to wait for 'non-proxy role' containers without healthchecks to stay in the running state readiness_delay: 10"

This change fixes two things:

1. Fix search for some terms

Lunr does do substring matching, but only with the end of a string (source)

So searching for deploy_timeout didn't match any pages because it was indexed as bootdeploy_timeout instead of boot deploy_timeout (see content strings above).

2. Fix previews in search page

Before

After

[djmb]

Brilliant, thank you @3v0k4 👍