Sitebeam 5.1

This version was released on 5th November 2012.

Summary

  • Vastly improved PDF exports.
  • Automated correction of most untestable websites.
  • Reports now automatically re-run themselves if they fail.
  • Completely new help area.

Detailed list

Improvements

  • Completely rewritten PDF exports.
    • PDFs no longer require a plugin or Google Chrome to use.
    • They are significantly more attractive, with better layout.
    • You can now export comparison reports.
    • Support for A4 and Letter paper sizes.
    • PDFs can be accessed via the API (selected customers only).
  • Added ability to test sites from different physical locations: UK, US, Netherlands. This makes it possible to see sites which deny or control access when viewed from different physical locations.
  • Entirely new report queuing mechanism allows for reports to be queued even if no Test Servers are available, and allows them to automatically restart if they fail (up to 3 times, before aborting). For example, if Sitebeam is being upgraded or there is Amazon downtime, reports will re-queue themselves up and explain the situation to the user.
  • Added detection of Clicky analytics.
  • Added detection of Google Tag Manager analytics.
  • Where the name of a website is specified in Open Graph tags (og:site_name), that name is used automatically to label the site instead of the web address.
  • Sitebeam now cleans up excluded words that have since been added to the dictionary.
  • Added ability to add a site to Sitebeam even if the site can’t be read during the adding process. Allows for sites that need advanced options, like a custom location or similar to be added.
  • Sitebeam is now smarter at solving certain common problems, and will do so itself where possible:
    • If the spider finds no pages, “Allow duplicate URLs” is checked and the site is re-spidered automatically.
    • Sites which appear to have no links in the first few pages are now spidered with Javascript scanning enabled automatically.
    • Sites which combined a mix of www and non www prefixes, but which only contain the homepage without the www prefix now automatically test.
    • Sites which return only 1-2 pages and have a suspected cause are stopped before their reports are viewed (non-Viewer users only) to explain; the user is given the chance to add extra domains to help test the site properly.
    • Sites which are embedded solely within a frameset to another domain are now considered to automatically include the target domain(s). When adding them, Sitebeam explains they are not really the website they meant, and corrects the web address automatically.
    • Sites now automatically retain cookies set by early pages even when duplicate URLs are disabled. This addresses many common spidering issues with sites that redirect back to set a cookie.
    • Stopped the spider from denying repeats of the homepage by default. This behaviour (whilst undesirable) was sometimes used by sites to set a cookie and redirect back to themselves; this is now acceptable as long as the “Allow duplicate URLs” box is checked.
    • When no pages are found, the spider narration now shows the list of all events found, instead of the list of zero pages found automatically.
    • When only a few pages are found, Sitebeam logs in detail the pages which are refused, aiding troubleshooting if insufficient pages are discovered.
    • Added option to test sites that return HTTP errors (e.g. 404/500) for all their legitimate pages.
    • Removed some superfluous spidering options as these are now automated.
  • Changed way expired accounts work: users can now log in, but are unable to run new reports. They can however view existing reports.
  • The Test URL feature now also displays any links found by Sitebeam.
  • Test servers now recognise and recover from crashes much more quickly.
  • Added support for spidering sites on non-standard ports.
  • The title of the page/tab changes to show the number of running reports. This helps when running reports in the background.
  • Slightly polished some CSS styles and transitions.
  • Upgrades are now much faster (approximately 4 seconds each; previously around 1 minute). This allows all customers to be upgraded much more quickly and with less downtime.
  • The summary status screen (i.e. clicking on “x running” in the top right corner) now shows the percentage of a report that is complete, where known.
  • New Favicon, which changes for Silktide users logged into Test Servers.
  • The spider is now more intelligent at managing how many processes it uses; it is less inclined to start or cancel spider processes.
  • For Silktide only – added ability to log HTML downloaded per-page in spider trace (check the “Debug trace” option).
  • For Silktide only – added ability to restart or abort all reports running on a specific customer install.
  • Added more explanations to the Admin > Test screen.

Changes

  • Disabled the Internal Links, Outgoing Links and Metadata tests by default. These are  informational only tests that do not contribute to scores; our usage figures show they was rarely used and occupied considerable space when printed. You can re-enable either under Account > Tests.
  • Completely removed the Spiderability Test. The latest Sitebeam spider is incompatible with the approach used by this test, as it tends to spider sites that were previously impossible automatically. Similarly, Google and other search engines are now known to spider previously unspiderable sites in most cases.
  • When adding a site, Sitebeam now follows up to 3 redirections (previously 2) to attempt to add the site.
  • Sitebeam no longer sends a HTTP Expect header in requests. This generally makes downloads faster and Sitebeam less liable to be flagged as a bot.
  • The Export button is now labelled the PDF button, when viewing a report.
  • The Content summary now penalises the absence of content less than before for small sites. The description of the summary now explains that this score represents the quality and volume of content.
  • The Content keywords test now considers keywords which are not present at all in the content against your score (previously, these were ignored – but this lead to confusion and possible abuse where ludicrous keywords were included).
  • The Broken links test now excludes HTTP 401 responses – these are not considered broken.
  • The Broken links test now reports all links in a simpler format (it doesn’t report different totals for possible and definitely broken links, which was sometimes confusing).
  • Reduced size of client logo on online reports, to ensure layout looks better for most users.
  • Moved Support button out-of-the-way on the test screen.

Fixes

  • Made it impossible for users to exclude the Languages test. The Languages test is required for the Spelling and Readability tests to function. Previously if a user disable Languages they would cause their own reports to fail.
  • Fixed flaw in Speed and Broken Files tests which caused them to run less quickly than possible.
  • Rewrote CSS parser to avoid potential crash when a page nested an exceptionally high number of unclosed tags within each other (higher than the stack limit of the operating system).
  • Made the W3C compliance check handle the official W3C parser breaking and returning a malformed XML response, as appears to occur for some sites with invalid document types.
  • Fixed a bug where Test Servers would fail to clean up after a report was run, making an eventual crash (from exhausted disk space) guaranteed.
  • Fixed some miscellaneous memory leaks and performance bottlenecks (in particular: the process log was committing excessively quickly during some tight loops).
  • Facebook detection now recognises new-format Facebook Like buttons pointing to a page, and Like Boxes (HTML5 variant).
  • Fixed issue where some Administrator accounts created by another Administrator that was subsequently deleted would be reset back to Builder accounts following an upgrade.
  • Fixed issue where malformed HTTP response codes could cause unpredictable behavior.
  • Fixed issue where updated Twitter API was incompatible with Twitter test.
  • Fixed W3C compliance test not recognising pages which return a HTTP error as having errors.
  • (Silktide only) Fixed bug where a user submitting some forms of feedback could submit a corrupted subject.
  • Fixed some minor typos, refined some text explanations.

Was this article helpful? Contact our support team if you have a question.