Expected Behaviour

I expect to land on a page with some relevant information when I click on:

Actual Behaviour

When clicking on the above list item I get a 403 forbidden page:

Reproduce Scenario (including but not limited to):

Open Google Chrome and go to: https://www.project-helix.io/pipeline/README.html.

Platform and Version

Google Chrome Version 90.0.4430.85 (Official Build) (x86_64) - MacOS BigSur

to support hlx installation via curl:

curl -sL http://www.project-helix.io/cli.sh | sh

We should be clear on who is going to look at project-helix.io and what kind of info they expect to get vs what kind of info we want to share.

Is having git installed really a pre-requisite to get started? i haven't tried without but i am not sure it really is...

I think at this stage MacOSX is a pre-requisite though, at least for the 60s version.

In order for the site to be properly indexed and the search feature to work as expected with need:

• robot.txt
• sitemap.xml referencing all the pages - that's where the subdomain trick (client, pipeline) will not work anymore).
  In a first step, the sitemap will be managed manually. In a second step it could be generated out of SUMMARY.md (but what about content that does not have an entry point here...).

This would also mean that the content is properly defined.

This should end with master-- not master--html. @tripodsan do you remember why you've made that change?

project-helix.io is supposed to be the entry point for helix documentation. Some parts like the hypermedia-pipeline or helix-cli are stored on various git repo: they all have their own documentation (which might be Helix agnostic) which we need to referenced from the project-helix.io.

Idea is to use multiple namespaces (strains / fastly services), like:

• www.project-helix.io (code: project-helix.io, content: project-helix.io)
• hypermedia-pipeline.project-helix.io (code: project-helix.io, content: hypermedia-pipeline)
• helix-cli.project-helix.io (code: project-helix.io, content: helix-cli)

The entry points to those "subdomains" would be handled via a hand-managed navigation, centralised inside a SUMMARY.md stored in the project-helix.io repo.

Goal here is to create the necessary structure to handle those "subdomains", add navigation references to those which have relevant content and to document how to onboard a new subdomain.

During the last outage, when one of the actions was missing,
fastly didn't send proper 500 page.

eg: see https://www.project-helix.io/noexist.html

With test/smoke/test.hlx.up.js, there's already a smoke test for hlx up i.e. the dev environment. We should extend it to also test hlx deploy and publish in order to ensure the integrity of deployed action and the published site.

We should inspire from sites or templates like git-tower.com, stripe.com, developers.squarespace.com and / or Bootstrap Grayscale

Expected outcomes: mocks or even better a css and html template

the currently specified circleci/node:8-stretch (debian 9) causes build failures:

Running smoke test using: hlx
libcrypto.so.1.0.0: cannot open shared object file: No such file or directory

see also adobe/helix-cli#317

Once the domain name has been registered (see #1),
we need to configure the Fastly service.

• Create the services in Fastly following the documentation https://github.com/adobe/project-helix/blob/master/getting-started.md#onboarding-a-domain
  • Create a service for project-helix.io
  • Create a service or strains for www, hypermedia-pipeline, helix-cli subdomains
• Update the documentation if needed

Proper linting setup should be added to the project to have nice looking code and avoid complains from LGTM.

• either with a fastly anycast IP
• or maybe directly on Google Domains

The schema documentation right now links to the raw files on GitHub instead of being served through Helix, because the foo.schema.html is interpreted as a call to schema_html.htl, i.e. schema is the selector.

I've been thinking about approaches to solving this including:

1. getting jsonschema2md to generate files that don't have more than one dot in the file name
2. allowing a configuration in Helix to determine which names are valid selector names
3. allowing Helix to do fall-backs in case a selector function can't be found
4. finding a project-level workaround

As I think configurations (2) and fallbacks (3) complicate things, the most promising options seem to be (1) or (4). One way to implement (4):

• create a schema_html.htl and schema_html.pre.js
• in pre.js add a before.fetch extension point that replaces foo.md with foo.schema.md
• go on as before

Following up on adobe/helix-cli#614

The submodules and the strains where built to support multiple content repository and local development: the local checkout of the submodules allows to navigate locally to those websites. I think the helix-cli supports now this use case via the strains and the urls. We could get rid then of the submodules.

When i started to work on helix it was hard to find any information how to create composed page that has different scripts for navigation, footer etc. There is short mentioning about ESI tags in the cache section. It would help a lot to have some dedicated point with an example.

The commit 3dda31c changed subdomains/hypermedia-pipeline to subdomains/helix-pipeline in the .gitmodules file but didn't update the entry in the subdomains directory.

This means that running git submodule init errors with the output:

fatal: No url found for submodule path 'subdomains/hypermedia-pipeline' in .gitmodules

Go Live of the project-helix.io site with a minimum of content, helpx styles and some links leading to external repo (hypermedia-pipeline and / or helix-cli) documentation.

Go Live pre-requisites is a deep content review!

We need to keep the documentation up-to-date regarding the debugging capabilities:

• how to debug HTL
• how to debug the pre.js
• how to debug with Chrome
• how to debug with VSC plugin
• how to console log the payload
• ... ?

This is a minor thing but might be worth to mention. When i try to debug helix demo project on windows using vscode, the instruction doesn't work. I need to add remote server or debug the process manually.

https://www.project-helix.io/doc/getting-started/how-helix-works.html

The flow has changed to include a singledispatch action in order to reduce the number of requests between fastly and runtime. This should be reflected in the diagram.

An Helix project comes with some CI capabilities. We need to make sure here they are relevant and are on the radar. The goals here are:

• make sure the project-helix.io is always running and accessible
• project-helix.io code and content is tested against the latest Helix code
• we have a Go Live procedure which makes sure everything is tested and nothing is broken

Following the instructions on the README.md to deploy and publish the site, hlx publish fails with the following output:

$ hlx publish
Publishing [==------------------------------------------------] preparing fastly transaction 0.0s
error: Error while running the Publish command: FastlyError: Token XXXXXXXXXXX expired at 2019-04-16T08:19:38Z
    at request.then (/Users/stefan/.nvm/versions/node/v10.15.3/lib/node_modules/@adobe/helix-cli/node_modules/@adobe/fastly-native-promises/src/httpclient.js:100:17)
    at process._tickCallback (internal/process/next_tick.js:68:7)Token 2KGK5kEqD7lAuJbumSPjcl expired at 2019-04-16T08:19:38Z
Error while running the Publish command

The github app needs a nice domain (and not adobeioruntime.net).

suggest:

• create a new fastly service for app.project-helix.io
• create basic backend to point to: https://adobeioruntime.net/api/v1/web/tripod/default/helix-github-app

@trieloff ?

For a quick startup, we will just re-use helix-helpx look&feel and code, i.e. css, htl and pre.js.
Since project-helix.io is a reference project, we only need to make sure this still follows the current best practices (helpx is 2 or 3 months old)

Open https://www.project-helix.io/index.html
-> Nav

Open https://www.project-helix.io/
-> No nav!

The header showing who modified the content last and when shows "invalid date"

e.g. https://www.project-helix.io/doc/getting-started/how-helix-works.html

In order to create meaningful content, we should have a clear picture of the content we want to output. The more precise we are on the sitemap, the better structured and quality content we will have.

I couldn't find in the documentation any information how to create a page with dynamic elements. For example if it is possible or how to create dynamically computed navigation.
It can be created using github api in the rendering script but does helix provide any support for this? Github api has limits and they might be low for anonymous user, this could be resolved by account support.

It would be great to have some concept outlined in Documentation. If it is not recommended to build such structures with helix it is good to mention it there also.

The homepage should introduce to project helix and some initial content should be inspired / copied from https://github.com/adobe/project-helix.
The dev quickstart is probably the most important piece of content to import here.
A navigation stored into a SUMMARY.md file will also be necessary to orchestrate the content navigation.

λ curl -sL https://www.project-helix.io/cli.sh | sh

downloading hlx installer v0.13.10 ...

% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 605 0 605 0 0 605 0 --:--:-- --:--:-- --:--:-- 860
100 36.2M 100 36.2M 0 0 7414k 0 0:00:05 0:00:05 --:--:-- 8659k

running hlx installer v0.13.10 ...

Verifying archive integrity... 100% All good.
Uncompressing SFX installer for hlx 100%

cp: cannot create regular file '/usr/local/bin/hlx': No such file or directory
failed to install hlx, aborting

Part of the overall work: adobe/helix-home#140

Smoke tests are currently broken because of the expectation to find a main branch here. See https://app.circleci.com/pipelines/github/adobe/helix-continuous/4130/workflows/8897b2a4-4e23-40cc-af69-bae43780bacf/jobs/19453

maybe we need to rewrite the helix-config.yaml to use conditions?

@dominique-pfister @trieloff ?

payload.content.children is deprecated but still in use in the project. I must be removed.

Register the project-helix.io domain name to the appropriate authority.

Would you be doing that ? @davidnuescheler @trieloff ?

Description

The error of the action invocation should not be shown to the end user.
we should probably have a custom error page in fastly.

Search box is there but does nothing.
Goal: use google search on the project-helix.io site.

Run hlx up
-> all requests /assets/*.png returns 404.

hlx version: v0.10.10

problem can be seen on www.project-helix.io.
if a X-Strain cookie is set, the ESI include for the summary also tries to load the summary from client.project-helix.io, although it the esi include is for www.project-helix.io

<div class="main">
  <div class="nav">
  <!-- include https&#x3a;&#x2f;&#x2f;www.project-helix.io&#x2f;SUMMARY.summary.html-->
  Error 404: 404: Not Found
   
  </div>
  <div class="content">

Action Logs:

    "2018-11-20T02:39:40.135435024Z stdout: error: Could not fetch Markdown from https://raw.githubusercontent.com/adobe/helix-cli/master/SUMMARY.md"

Cookie:

After clearing the cookie, the ESI include works. but then the cookie is set again...

After upgrade to helix-cli v0.9.1, the site renders partially:

• Nav (https://www.project-helix.io/SUMMARY.summary.html) renders a white page
• Index (https://www.project-helix.io/) renders only the title and commits history

This happens only on production, works fine with hlx up

Just run hlx deploy in the project (after decrypting and sourcing the env file) - you get error the following error:

[==================================================] analyzing html.js 0.0s
[==================================================] packaging zwitch 0.0s✅  packaging completed
[====================------------------------------] deploying summary_html.js 0.0sPUT https://adobeioruntime.net/api/v1/namespaces/helix/packages/helix-services?overwrite=true Returned HTTP 409 (Conflict) --> "Resource is a package and cannot be converted into a binding."
warn: None of the strains contains package information. Aborting command.

@tripodsan feedback: use hlx deploy --static=build instead.

I think hlx deploy should work with any extra magic property.