* master:
Adding Mother Jones feeds for testing.
Fixing the ghost story unread count bug. Turns out it is a weird condition where a story changes its date after being published, but a user's mark_feed_as_read date is inbetween the new date and the original date, leading to a ghost unread story that can't be read.
Fixing errant linkify text bug that resulted in some javascript being executed and overwriting the page. Thanks to @aaronlidman for the help in identifying the bug.
Conflicts:
assets.yml
* master:
Adding tests in prep for Mother Jones duping.
Fixing rss feed tests.
Adding press story from Italian site Skimbu.
Adding Cuk.ch story about NewsBlur.
* master:
Fixing issue around feeds that are mistakenly declared as not having a page.
New audio control, with Flash fallback for browsers like Firefox/IE.
Conflicts:
assets.yml
* master:
Allow login from signup form.
Reducing celery prefetch to get more accurate queue sizes.
Removing RabbitMQ dependency. Moving to redis for task queuing.
Counting queue size for celery/redis.
Switching to redis to feed queuing. Also adding a weight for skipping feed fetches under high load.
Skipping feed resurrection.
Fixing last update problem that has been choking feed fetchers.
Fixing last update problem that has been choking feed fetchers.
Typo in feed pool drain.
Adding ability to wind down feed queue by faking requests.
Adding debug message about counting errors.
Conflicts:
fabfile.py
* master:
Fixing starred stories from not loading past the second page due to a missing flag reset.
Fixing user-found bug around tooltips continuing to show if the original dom element (and the reference to the tooltip) are destroyed too early, leaving a tooltip floating in the air.
Geometrically delaying the fetching of bad feeds, while they're still somewhat fresh.
Fixing error codes on statistics. Also fixing simple error on invalid feed.
Update media/js/newsblur/reader.js
Conflicts:
media/js/newsblur/reader/reader.js
* master:
Adding asset-less deployment.
Upgrading feedparser from 5.0.1 to 5.1.
Catching errors in page links with ill-formed urls.
Not showing the world favicon until favicons are downloaded.
Fixing feed exception dialog to actually reload feeds on the page.
Geometrically decaying feed fetches on known good feeds that have gone bad.
Fixing issue around empty original pages.
Conflicts:
media/js/newsblur/reader/reader.js
* master:
Re-styling the crap out of all taskbar buttons. Much more streamlined now.
Tweaking the living crap out of the intelligence control's colors. Lightened.
User feedback: hiding positive and negative unread counts in title if there are no unread stories outside of neutral.
Correcting typo in dupe feed resolution.
Local settings in fabfile.
Fixing up a minor detail in the new intelligence control.
Rewriting intelligence slider to be an intelligence control (segmented control). Should make a ton more intuitive sense.
Trying to fix IE troubles by forcing single requests per read story.
Fixing nginx conf to support ssl.
Adding PID files to gitignore.
Ignoring files.
Adding certificates to ignore.
Adding https to nginx.
Conflicts:
media/css/reader.css
media/js/newsblur/common/assetmodel.js
* master:
Adding build/ back to .gitignore.
Updating .gitignore.
Using dsa pem for retrieving rabbit update counts.
Preserving feed fetch error histories for 2 weeks.
Adding munin graph for known good feeds.
Turning down verbosity on feed tasking.
Adding known good feeds to munin to chart their progress.
Fixing headers overflow in Chrome.
Adding new feeds db column 'known_good', for forcing updates of feeds that were once good but are now throwing 500s.
Adding blogging OT press story.
* master:
Fixing broken opml export due to missing feed icons.
Fixing slight bug in how active premiums were applied to feed update times.
Shift+U switches between read and unread. Thanks to @MLACitation: https://twitter.com/mlacitation/status/155371014560288768
Adding active_premium count. Setting a minimum of 1 hour fetches for feeds with any active premiums.
Allowing username refinement in fetching starred stories.
Increasing timeout on gunicorn to 2 min.
Adding management command to fetch starred stories on command.
* 'social' of github.com:samuelclay/NewsBlur:
Adding error handling for insta-fetching stories.
Turning off the page fetcher requests vs. urllib2 discrepency checker. I bet I've caught them all by now.
The @kennethreitz commit: logging all discrepencies between requests and urllib2. This should take a few hours.
* master:
Abort old requests and don't cache responses. Trying to fix IE read story bug.
Force jquery to never cache.
Checking for exceptions earlier in order to show fixed site.
Changing user agent to hopefully override overly-sniffy websites.
Fixing inaccurate substitution on feed urls. Ugh, can't believe this was here this long.
Adding the setting of a feed's view settings to Site Settings dialog. This should've been in there a loooong time ago.
* master:
Fixing feed loads only once for those with skipping stories.
Adding handler to to_json calls on all serialized models.
Fixing two major bugs: auto filling stories now works far better nad doesn't lock you in. Also fixing the infernal unicode json parsing bug that's been effecting a bunch of sites.
Adding WP7 as user agent.
Adding necessary import when not running from the REPL.
