Removed bootlint from bower-managed packages.
Bootlint no longer supports bower, in facts it includes a deliberately malformed bower.json to stop it from being imported with bower, see: https://github.com/twbs/bootlint/issues/330 Bootlint...
View ArticleBower component versions are explicitly specified in the Bowerfile.
This way a specific version is installed by bower, instead of just using "latest" version. This makes it much easier upgrading individual bower components, instead of having to run a "rake...
View ArticleAdd all necessary paths for bower components, so that the asset pipeline find...
The initialization code that adds the paths for each frontend package (bower modules) is now much more complex. Unfortunately every module has the necessary assets (js, css, scss etc files) in a...
View ArticleRemoved precompilation of individual js files.
Those files are included in the asset manifesto and will be included in the compiled asset bundle anyway. They will not be included from the html individually and it is not necessary to individually...
View ArticleRemoved fastclick initialization from javascript.
Fastclick was removed in commit 862e7ee8 but the app still tried to initialize it on each page load. This speeds up page load (probably imperceptibly, but still) and removes a js console error.
View ArticleUse rails helpers for the font paths in fontawesome SCSS, instead of hardcode...
The paths used by default by fontawesome do not exist in a rails app, the asset pipeline creates different paths. Using rails helpers is a must to find out the actual asset path for each font. There...
View ArticleReplace calls to ActiveSupport's "try" with the safe navigation operator "&."
The safe navigation operator is new in ruby 2.3.0 For more details see for example: http://mitrev.net/ruby/2015/11/13/the-operator-in-ruby/
View ArticleAdded Zlib::DataError to the errors that can be expected to be raised when fe...
If this (rare) error happens, this means that something is wrong with the zipping of the HTTP response. The error is not raised, to avoid sidekiq adding the worker to its list of retries; rather, the...
View ArticleUse a fork of rack-cache.
This fork allows setting the entity store to nil. When configured this way, it returns empty bodies for cached responses, instead of returning the original body. It doesn't persist entities, which...
View ArticleWhen fetching feeds using rack-cache for client-side caching, set entitystore...
This keeps rack-cache from persisting response bodies in disk. We are not interested in those bodies, if a response indicates that the cache for that feed is still valid, all feed processing can be...
View ArticleSet Rack-cache verbosity to false.
This removes some messages from the logs. Also will probably remove IO errors that happen occasionally when trying to write logs from rack-cache.
View ArticleRun FixSchedulesWorker daily at 6AM instead of 5AM.
This way this scheduling does not clash with the automated sidekiq restart that happens at 5AM. This fixes a bug: FixSchedulesWorker was often not running at the scheduled time.
View ArticleMerge pull request #13 from yogodoshi/update-gems-with-vulnerabilities
Update nokogiri and devise gems because of vulnerabilities
View Article
