Permalink
Switch branches/tags
Nothing to show
Commits on Jul 31, 2015
  1. Add Dockerfile.ci and docker-compose.ci.yml

    grosskur committed Jul 31, 2015
    This provides a base image for CI systems to use.
Commits on May 4, 2014
  1. Bump requirements.txt to latest versions

    grosskur committed May 4, 2014
Commits on Feb 23, 2014
  1. Optimize "fork me" image

    grosskur committed Feb 23, 2014
    Run it through pngcrush and pngnq. Reduces it from 7791 bytes to 3776
    bytes (~50% reduction).
  2. Use gzip encoding in production

    grosskur committed Feb 23, 2014
    Enabled when DEBUG is not set.
Commits on Feb 22, 2014
  1. README.md: Tweak wording

    grosskur committed Feb 22, 2014
  2. Make collector run forever

    grosskur committed Feb 22, 2014
    This is more consistent how Procfile processes are supposed to
    behave. The old behavior is still available with the --once flag,
    which makes it suitable for running from Heroku Scheduler.
  3. Inline main.js code in template

    grosskur committed Feb 22, 2014
    main.js is awkward because it relies on the prices() function to be
    previously defined. So it didn't fit into the asset pipeline.
    
    And now that we support CDNs, we don't want to deal with cache
    invalidation for main.js. So just inline it in the main.html template.
  4. Clear config environment variables after reading them

    grosskur committed Feb 22, 2014
    We use environment variables for configuration. But they don't need to
    persist after application startup, so delete them.
  5. Allow serving static content from a CDN

    grosskur committed Feb 22, 2014
    This is controlled by STATIC_HOST. For example, set
    
        STATIC_HOST=mydomain.cloudfront.net.
  6. README.md: Don't recommend setting DEBUG=1

    grosskur committed Feb 22, 2014
    Asset compilation now works properly since we started using rjsmin.
    Also tweak the title.
  7. Freeze requirements.txt

    grosskur committed Feb 22, 2014
    This makes deployments more reproducible.
  8. Switch to alternate method (HTML tag) of Google site verification

    grosskur committed Feb 22, 2014
    This is cleaner than the default method (HTML file upload).
Commits on Feb 21, 2014
  1. Add support for Google Analytics

    grosskur committed Feb 21, 2014
    Use the environment variables GA_TRACKING_ID and GA_DOMAIN to
    configure this.
  2. Add support for a Google Webmaster Tools verification

    grosskur committed Feb 21, 2014
    Google Webmaster Tools wants you to serve a custom URL in order to
    verify ownership of your site. Set the environment variable
    GOOGLE_VERIFICATION_ID to enable this.
  3. Add favicon

    grosskur committed Feb 21, 2014
  4. Add /healthcheck handler

    grosskur committed Feb 21, 2014
    This is a handler that just prints "OK" and returns 200. It's useful
    for monitoring and load balancing setups.
Commits on Feb 19, 2014
  1. README.md: Update instructions to skip asset compilation

    grosskur committed Feb 19, 2014
    Temporarily recommend setting DEBUG=1 to skip minification of css and
    js files. Minified nvd3 seems to cause a problem.
  2. collector: Fix handling of initial run

    grosskur committed Feb 19, 2014
    Everything actually worked OK because arrow.get(None) returns the
    current timestamp. But make this clearer: If no previous run has been
    detected (i.e., no stored timestamp found) use an initial window of 1
    hour.
Commits on Feb 18, 2014
  1. README.md: Fix instructions to mention Heroku scheduler

    grosskur committed Feb 18, 2014
    It's more cost-effective to run the collector from the scheduler.
  2. Add post_compile script generate assets

    grosskur committed Feb 18, 2014
    Run 'webassets build' at the end of the build process.
  3. Switch database from Postgres to DynamoDB

    grosskur committed Feb 18, 2014
    The dataset is growing so let's try using DynamoDB. For now, we're
    using both boto (for DynamoDB operations) and botocore (for querying
    spot price data). Eventually, we'll move everything to botocore and
    possibly use the pynamodb interface. Also:
    
    * Start collecting data for all EC2 "products":
    
      * Linux/UNIX
      * Linux/UNIX (Amazon VPC)
      * SUSE Linux
      * SUSE Linux (Amazon VPC)
      * Windows
      * Windows (Amazon VPC)
    
      Previously, we only collected data for Linux/UNIX.
    
    * Handle next_token properly to pull spot price data that is more
      than one page.
    
    * Save the "end_time" timestamp of each run, so we know where to
      start from next time.
    
    * Update to latest bootstrap, jquery, and nvd3.
    
    * Tweak timestamp on graph X axis to include HH:mm.
    
    Performance-wise, the app still feels about the same as when loading
    data from Postgres. Possibly it's a little bit slower. This can be
    tuned in a future change. We might even use memcache to reduce the
    number of DynamoDB operations.
Commits on Dec 4, 2013
Commits on Apr 20, 2013
Commits on Apr 19, 2013
  1. Fix formatting in README.md

    grosskur committed Apr 19, 2013