3 ms·
I won't try to address every part of this writing, but there is one claim that I feel is false: namely that the AMP validation rules are not open source. Here
by gregable 5y ago
I won't try to address every part of this writing, but there is one claim that I feel is false: namely that the AMP validation rules are not open source.
Here is the code: https://github.com/ampproject/amphtml/tree/main/validator https://github.com/ampproject/amphtml/tree/main/validator
All of the rules are written in human-readable config language. Here are the rules for common HTML elements (https://github.com/ampproject/amphtml/blob/main/validator/validator-main.protoascii https://github.com/ampproject/amphtml/blob/main/validator/va...) and there are rules per-amp-tag (example: https://github.com/ampproject/amphtml/blob/main/extensions/amp-vimeo/validator-amp-vimeo.protoascii https://github.com/ampproject/amphtml/blob/main/extensions/a...)
You can build it, run it locally, etc. There are open-source implementations that run in the browser, in a vscode plugin, on the command line via NPM, etc. Other entities can and do use the same code for their validation of AMP. Pull Requests are accepted from anyone and those changes land in Google production and all of the aforementioned tooling targets within a few days. There is documentation on how to contribute to and edit these files (https://github.com/ampproject/amphtml/blob/main/docs/component-validator-rules.md https://github.com/ampproject/amphtml/blob/main/docs/compone...).
- dmitriid 5y ago> namely that the AMP validation rules are not open source. Here is the code Yes. That's the trick big corps play on you. They develop some code, and push it to GitHub. - all of the decisions where the code is going belong to the corp - all (or at the absolute vast majority) of contributions come from the corp's developers - there's little to no insight into why things are the way the are. Sometimes you can be big enough to gain acces, or important enough. Before "open governance" it was some publisher group set up by Google that was supposedly there to monitor But sure, it's open source. > Pull Requests are accepted from anyone and those changes land in Google production 1. What's the ratio of PRs from Google employees to those from "anyone"? 2. Indeed. They land in Google production. The company with $183 billion in yearly revenue thanks you for your unpaid contributions (if they ever make it) --- BTW, there are 1 500 open issues in that repo. Quite a lot for some validation rules.
- gregable 5y agoI did some hacky analysis and roughly half of the PRs originated in GitHub and the majority of those are not google employees. Most changes have a github issue attached with discussion, all of the changes have descriptive commit messages regardless of who authored them and there is significant effort to document the decisions and decision process throughout. I can't speak for the release frequency for other organizations' production that consumes this code. That includes many publishers, as well as other AMP caches such as Bing and webmail clients like Yahoo. The repo is for all of AMP, not just the validator. The validator related issues are handled by the "caching" working group (since caches are the consumer of the validator) and you can search for them with this query: https://github.com/ampproject/amphtml/issues?q=is%3Aissue+is%3Aopen+sort%3Aupdated-desc+label%3A%22WG%3A+caching%22 https://github.com/ampproject/amphtml/issues?q=is%3Aissue+is...
- dmitriid 5y ago> PRs originated in GitHub and the majority of those are not google employees. How did you figure that out? There are people don't list their company, but their only activity on GitHub is contributing to AMP. Some people don't list Google, but are a part of the AMP org on GitHub. > all of the changes have descriptive commit messages Please show me all the open decisions that introduced AMP for email. (hint: you can't) As I mentioned in a sibling comment: why are people so surprised that it's Google's product driven by Google's own needs?
- jsnell 5y agoYou're simultaneously complaining about how all the commits are coming from the company and that they're exploiting anyone making commits. You can't have it both ways, pick one. The repository has 1k contributors over 20k commits. That seems like an exceptionally diverse contributor base.
- dmitriid 5y ago> The repository has 1k contributors over 20k commits. That seems like an exceptionally diverse contributor base. Go to the contributor tab and look at main contributors. It's entirely a Google-driven product. Why is this a surprise to anyone? Because it's on GitHub?
- tyingq 5y agoThe real important part of AMP that isn't open source is this, from the AMP HTML Specification[1]: Required markup AMP HTML documents MUST ...(some deleted) - contain a <script async src="https://cdn.ampproject.org/v0.js"></script> tag inside their head tag. That's the magic that allows google to inject the AMP header onto the top of your page, hijack the back button / left-right swipe on your page, etc, when someone visits your page from the carousel. It's hard to call something open source when the spec says "You must include this javascript url (and therefore, code) that you don't control". And the validator does seem to make sure it's there. [1] https://amp.dev/documentation/guides-and-tutorials/learn/spec/amphtml/ https://amp.dev/documentation/guides-and-tutorials/learn/spe...