Hi, I have done some work on the infrastructure behind the website recently; Here is a highlight of the changes:
1) new workflow: the staging website is back - pushes to https://github.com/freenet/website/tree/2016-redesign get deployed by travis (https://travis-ci.org/freenet/website) on http://staging.freenetproject.org/ - tags get deployed live through the same process on https://freenetproject.org/ workflow-wise, it's very similar to what we used to have (except previously it was two branches)... we could always revert back to master/next if someone can come up with the magic git incantations required. 2) HTTP headers are now set like they were We are now setting a bunch of security headers on top of the cache headers. I have decided that maintaining the cache attributes on the S3 objects wasn't a good way of dealing with it going forward... so now there is a lambda function that hooks onto "origin requests" and decides what headers to set depending on the path of the file requested. I will commit it to the repository shortly. In practice, that currently means that we cache content (both browser- side and on the CDN) only in production, 500 seconds for HTML and 90 days for everything else. The toolkit we use (webassets) makes it easy to generate content-dependent URLs that will invalidate the cache when required. 3) l10n I have now re-enabled l10n (french only so far) and coded another lambda function to auto-select the right translation based on the request from the user (it parses the Accept-Language header). It works for me and I think that we could get rid of the UI components of it. The next step should be updating/removing/curing the actual content, then we can ask translators to update their translations and re-enable them. Do we have a native English speaker who can volunteer for this? Florent
signature.asc
Description: This is a digitally signed message part