Include saucelabs gems only in bundle only in Travis CI.
Otherwise just having sauce-connect in the bundle demands having env variables with the username and API key for saucelabs, which makes no sense if we're not conna use the saucelabs cloud.
View ArticleSpeed up test by looking at some of the entries that have to be present in th...
Speed up test by looking at some of the entries that have to be present in the page, instead of checking all of them. This particular test is painfully slow (more than 5 minutes!). This commit makes it...
View ArticleAnother approach to have saucelabs gems in the bundle without causing problem...
Another approach to have saucelabs gems in the bundle without causing problems in environments other than travis ci.
View ArticleFix reporting to coveralls from the CI environment.
Also the rake task to run tests is now "rake spec:ci" instead of "rake spec:travisci". We've switched from travis to semaphore.
View ArticleFix error that caused & chars in feed and entry URLs become escaped into &
This was caused by the sanitizer, which assumes input is HTML and escapes any entities it finds. This behavior cannot be avoided, so the output from the sanitizer is passed to CGI.unescapeHTML() so...
View ArticleDo not escape HTML entities (e.g. & characters) in feed URLs when sending the...
Do not escape HTML entities (e.g. & characters) in feed URLs when sending them to the client as JSON. This fixes a problem where feeds with "&" characters in their URL had the wrong URL in...
View ArticleWhen updating a feed, if sanitization would turn the title into a blank...
When updating a feed, if sanitization would turn the title into a blank string, keep the old value instead. This is intended for feeds which have e.g. something that looks like an HTML comment for a...
View ArticleWhen autodiscovering a feed, if the autodiscovered url is relative and does...
When autodiscovering a feed, if the autodiscovered url is relative and does not begin with '/', prepend a '/' before converting it to an absolute url. This fixes error autodiscovering feeds with...
View ArticleAdded "published" attribute to EntryState model.
The idea is to denormalize the database so that the queries that retrieve paginated entries only have to follow the EntryState table and indexes, resulting in faster response time.
View ArticleAdded not null validation to "published" attribute of Entry model.
This should have been added long ago, it somehow slipped my radar.
View ArticleRemoved test, a test for the default value of "published" attribute of Entry...
Removed test, a test for the default value of "published" attribute of Entry model was already implemented.
View ArticleRepeat failed tests up to 3 times in CI environment only. Other environments...
Repeat failed tests up to 3 times in CI environment only. Other environments do not repeat failed tests. Trying to debug a failed test that is repeated is annoying.
View ArticlePublished attribute of EntryState model has its value fixed to the "published...
Published attribute of EntryState model has its value fixed to the "published" attribute of the associated Entry instance. No matter what value other code tries to set, it is guaranteed that an...
View ArticleAdded index to the EntryState table.
It will be used when retrieved a page of entries, thank to the denormalization of including the published attribute in the EntryState model.
View ArticleWhen retrieving paginated entries, set SQL order conditions on entry_states t...
This should help the database choose an execution query that gets results from entry_states as soon as possible, avoiding nested loops with entries. For the moment the entry creation datetime is no...
View ArticleAdded "entry_created_at" attribute to EntryState model.
Also included it in the entry_states table index which I expect will be used when retrieving paginated entries.
View ArticleAdded validations, default value and unit test for the entry_created_at...
Added validations, default value and unit test for the entry_created_at attribute of the EntryState model.
View ArticleUse the EntryState entry_created_at attribute to order entries, when...
Use the EntryState entry_created_at attribute to order entries, when retrieving unread entries from a feed or folder (including the "all subscriptions" special folder). The ordering priority is: -...
View ArticleUse find_each instead of each to update all EntryState instances during migra...
Trying to use "EntryState.all.each" means rails attempts to instantiate all EntryState objects, which will give an outofmemory error and the kernel will kill the ruby process. Using...
View ArticleFixed migrations.
The migrations I'm trying to apply use the EntryState model. In its current state it makes reference to the new "published" and "entry_created_at" attributes. This means that for the deployment to run...
View ArticleFix migration.
The current value for the already existing records in entry_states table for the new published and entry_created_at columns is set with a raw SQL update, instead of using ActiveRecord. Using AR is too...
View ArticlePerformance improvement when retrieving unread entries.
When retrieving unread entries, we already know that all entries in the list will be unread (it is guaranteed by the queries that get them from the DB). Therefore the step of looking at entry_states...
View ArticleIncluded aws-sdk-rails gem, which pulls v2 of aws-sdk.
Version 2 of aws-sdk introduces changes that break backward compatibility, so some changes will be necessary.
View Article