More accurate entry hashing
The hash that uniquely identifies an entry in the scope of a feed is now calculated using the concatenation of its content (if present), summary (if present) and title (which is always present). This...
View ArticleCorrect order when deleting duplicate entries in migration
Last implemented rails migration deletes duplicate entries in a single feed based on title+summary+content md5 hash, not just guid. The oldest entry for each set of duplicates in a feed is left in the...
View ArticleFixed log generation
unique_hash attribute of DeletedEntry model may be nil, we have to guard against this when generating log messages.
View ArticleWIP: validate that new entry is not already deleted using unique_hash
This complements checking for duplicate guids in deleted entries when creating a new entry.
View ArticleUse id to order entries in migration
When deleting entries with duplicate hashes in this migration it is important to leave the oldest entry alone and delete any newer duplicates. The id attribute is used in the case of duplicate entries...
View ArticleFix: allow multiple deleted entries with nil hash for the same feed
Also fix a test that should be testing for this but wasn't actually testing anything.
View ArticleFixed migration
Now it doesn't invoke activerecord validations or callbacks, which can fail if the model class is out of sync with the db schema when running the migration. Instead we use methods that update the db...
View ArticleFixed error in migration
The wrong model was being deleted. Fortunately the migration failed because it generated the wrong SQL when trying to delete records from deleted_entries; the correct thing to do is delete from...
View ArticleRewritten migrations to update db indexes
The addition of the unique_hash column to entries and deleted_entries tables means it's a good idea to drop a couple old db indexes and create new ones adding the unique_hash column in each one. The...
View ArticleUpdated comments
Old comments mentioned entries and deleted_entries having unique guids in the scope of a feed, now they also have unique unique_hashes.
View ArticleAssociate validation errors to base object, not a specific attribute
When trying to save a new deleted_entries instance a validation checks that there isn't a non-deleted entry with the same feed_id and either guid of unique_hash. Similarly when trying to save a new...
View ArticleEarlier detection of entries with duplicate content during fetching
Until now the detection of entries with duplicated content (based on the md5 hash of the concatenation of content, summary and title) happened in an ActiveRecord validation just before saving each...
View ArticleGenerate favicon for iOS touch devices
It should get rid of many erros in production logs, when apple devices try to get a different icon than what's declared for HTML5 clients.
View ArticleFix to validation errors when fetching feeds
When processing entries in a just fetched feed, we build new entries with ```feed.entries.build``` before knowing if we actually are going to save the entry in the database. It's possible that the...
View ArticleFix error processing images in entries
If the img src is already the "loading" animated gif, do not process the image.
View ArticleShow entry validation errors in the log
When an invalid entry is fetched, show the specific validation error in the log instead of a generic "entries is invalid" error for the feed object.
View ArticleChange log message and level
When the validation that checks that an entry is not already deleted fails, the right thing to do is to add the error to the entry.errors list and write a log message with low priority. The class that...
View ArticleEnable following redirects in RestClient.
RestClient should follow redirects automatically, but I'm getting errors when a server returns a 301 code. It's probably caused by the rest-client-components gem, which I need for client-side HTTP...
View Article