SiteRay 4.0

From SiteRay wiki

Jump to: navigation, search

SiteRay 4.0 is currently under development. All details on this page may change.

Improvements

  • Added new (vastly simplified) progress meter for monitoring test status. The progress meter now indicates where the current report is in a queue.
  • Users can now abort running tests at any point.
  • Added new 'summary' reports, which run faster, use less space and contain much less detail (intended for tasks like pitching a new website). Some informational-only tests are excluded for summary reports.
  • Radically improved UI, faster and simpler throughout.
  • Report presentation revised, with a cleaner simpler style throughout.
  • Added new Recommendations section to reports, which summarises actionable recommendations for every test, ordered by relative priority and divided by role (developer, marketer, executive).
  • All reports have a new layout, with all tests visible at all times (together with their scores) on the left.
  • Sites can now be browsed visually using screenshots in a grid view (the old list view is still available).
  • New, simpler dedicated login screen, with greater support for custom branding.
  • Screenshots are much faster to generate, and always remain consistent.
  • When a screenshot is ready, the current page is updated automatically (no need to click "Refresh").
  • AJAX links in reports now work when being bookmarked or opened in a new tab. The back button therefore now works for nearly all intra-report links.
  • Site name is now automatic (based on URL), but can be overridden where desired.
  • New Help screen now displays recent, relevant articles from various blogs (including Silktide).
  • All AJAX pagination reworked to be easier to understand and use (including addition of "Prev" and "Next", removal of superfluous "Pages").
  • When selecting "Re-run" for a custom test, the previous test selections are now automatically selected.
  • Spider now calculates the position of links on each page which reference each other, allowing for contextual analysis of links (higher / lower in page). This information is now displayed in the Spider Narration view.
  • When viewing a report, the page title now includes the name of the website and type of report (previously was generic) - better for bookmarking and multiple tab usage.
  • Many tests now embed helpful feature boxes inside them, explaining key concepts, implications and recommendations.
  • Added totally new demo mode, where users can create their own user accounts and use SiteRay (with many restrictions).
  • When an install is new, or a user has no existing reports, they are automatically guided through creating their first website. Other areas (reports, dashboards etc) are still available, but disabled and explain that they should be returned to later.
  • Significantly revised Test Server UI.
  • Removed all support for Partitions (which were barely ever used), significantly improving overall performance.
  • Users can now login in on multiple computers simultaneously without being logged out elsewhere (note: excludes WebScan users).
  • When testing "www.example.com", SiteRay will now automatically assume you also want to include URLs in "example.com". To override this, uncheck the "Automate advanced settings?" box under site settings.
  • SiteRay now cleans up badly entered URLs automatically (e.g. normalising case, fixing bad formats).
  • Optimised overall AJAX and Javascript performance.
  • All core UI features now support friendly URLs.
  • Connections between UI and Test Servers now use a new communication and setup model: connections to a test server can be setup one way from a default config file, and connections back to a UI server automatically timeout after 1hr of non-use (essentially cleaning the pool of connections that are polled).

Fixes

  • Fixed various conditions whereby a report could become stuck mid-run, and would never automatically clear.
  • Fixed an issue where a test process being restarted mid-way through a test would cause duplicate results to appear within a test.
  • Fixed issue with the CSS parser where it would not always recognise selectors of the form "tag#id" correctly.
  • Fixed flaw where the calculation of plain text would sometimes incorrectly exclude text for pages with certain broken HTML formats.
  • Testing a site without a trailing end-slash no longer causes the screenshots to fail to generate.
  • Clicking Logout no longer sometimes causes a Javascript popup to appear saying "You have been logged out elsewhere" (the popup still appears, but not if you logout yourself).
  • Fixed issue where the internal count of links to a page would be thrown off if cookies were changes between consecutive instances of a page, yet cookie-sensitivity was disabled. This was more common that would be expected, as cookie-sensitivity is normally low.
  • Fixed Contact details test falsely detecting emails with different cases as different.
  • Fixed Broken lists test saying a broken link was (sometimes) found on the wrong page.
  • Fixed issue where new versions of Firefox would ignore the "no language" option, forcing selection of an undesired language automatically.
  • Fixed issue where URL chopping test would not always consider a 404 an acceptable form of URL handling.
  • Fixed issue where Flash parser would incorrectly suspect a Flash file called "expressinstall.swf" when certain code was used.
  • Fixed issue where URLs not correctly escaping a space character would not be handled as they would in a browser (only for some fringe cases).
  • Fixed performance issue where parsing certain HTML files could cause them to be necessarily re-parsed for a change in character set, when only the formatting of the character set had changed.
  • Fixed bug where removing a test would cause dashboards which used that test to crash.
  • Fixed bug where some parts of the CSS/HTML parser wouldn't correctly use a base URL when specified using the <base> tag.
  • Fixed issue where some non-HTTP schemes (e.g. "aim:") could be encoded into a URL, and would flag as broken links. SiteRay now detects these generically, rather than on a specifically excluded basis.
  • Fixed issue where mixing cases in the Speed (formerly File usage) test would confuse results for some specific file instances.

Other changes

  • Clicking on links in reports now opens the original page, not the Visual Inspector.
  • The Server admin screen has been simplified, many superfluous options removed.
  • The spider now recognises and excluses some newer tracking code URL query paramters, such as those used by Google / Feedproxy.
  • The spider now excludes pages which return an error code (e.g. 401, 404, 500 ...). The presence of broken links is still detected by the Broken links test, but now these broken links will not flag as additional errors for other tests (e.g. a missing meta description for a broken page).
  • The spider now excludes the relatively common "xxx/home.yyy" page as equivalent to "xxx/", just as before for "xxx/index.yyy" and other variants.
  • Home screen removed. Existing options have been moved under Settings or Help.
  • Score descriptions revised (e.g. "Adequate" is now "Average").
  • Admin screen is now called Settings, and available to all users.
  • Categories removed.
  • Partitions removed.
  • Global site search removed, moved into the Website tab.
  • Old "report view" moved from inside Websites tab into a new tab: Reports.
  • Layouts and report layout editing removed.
  • Developers no longer see a large panel of options when re-running a test. These are now either redundant or moved inside Site Settings where appropriate.
  • Many unnecessary features removed or purposely concealed.
  • The logic for considering a page as a probable redirect has been substantially overhauled to be more accurate for edge cases. This affects the accuracy of many tests on sites with badly behaved redirections.
  • Changed the default sort by order to "descending", when clicking on new columns in the AJAX table view.

Tests

New

  • Added new Visual interest test, which measures the visual interest in individual pages and the site as a whole, plus their variety. This in part replaces a portion of the old Images test, but uses totally new and vastly more sophisticated logic.
  • Added new Search Engine Placement test, which summarises Search Engine Result Pages (SERPs) based on Page Titles and Meta Descriptions. Uses similar scoring to previous Page titles test, but with more refinement and more up-to-date rules. In particular, the length, format, and repetition of titles and descriptions are now tested.
  • Added new Freshness test, which estimates how up-to-date a website is based upon dates written within it. This means a site can be dated, even if it has not been tested before.
  • Added new Writing test, which analyses the quality and style of writing within a website. This test has been trained on a corpus of over 1 million webpages to identify consistent language patterns.
  • The new Image usage test scores based on the use of explicit image sizes (as before) plus the use of web-friendly image formats, which was not tested before. It displays a list of all images in image tags used by the site (purposely excluding those in CSS, which it used to check before).
  • Added new Twitter test, which measures mentions of the website on Twitter, most influential tweeters, most popular tweets, and evaluates the use of any existing Twitter feeds which the website uses.

Removed

Existing

  • Many tests will now not display itemised detail when it is clearly irrelevant: for example, if a site has no Analytics solutions, the test will not list every page individually - the summary is sufficient to explain this.
  • All recommendations are now summarised in bullet form for individual tests, sorted in descending order of importance. You can view all recommendations for the whole website under the new Recommendations page.
  • All scores now use a scale from 0 to 10 (previously 1 to 10). This ensures that 5 is the exact average midpoint (previously, confusingly, it was 5.5). As a result, all scores will change from SiteRay 3. When upgrading old reports, all of these historical scores will use the new 0-10 scale automatically.
  • Analytics software test renamed to Analytics test.
  • Analytics test now supports these additional Analytics tools:
    • Quantcast
    • BBC (proprietary)
    • Google Analytics injected via jQuery
    • 103bees
    • Mint
  • Alternative text test now ignores img tags which don't have a src (i.e. empty images used in scripting).
  • Alternative text test now detects and excludes images of 1x1 pixel size and excludes these as tracking images.
  • Analytics test now almost excludes probable redirection pages (client side redirections) from being penalised.
  • Analytics test now scores overall proportionally due to page importance.
  • The Amount of content test now performs a much more granular analysis of page content, listing individual pages in turn and summarising them graphically.
  • The Amount of content test is now scored. Criteria include: is there enough textual content anywhere in the site, is it spread out well, are there many pages with negligible content on them.
  • The Incoming links test now breaks down links from Google, Yahoo and Bing. The scoring algorithm has been revised to take account of all three (previously, scoring was based exclusively on Google's links, aka 'quality links'). Clicking on a source now displays the latest links from that source.
  • The Broken links test now punishes large sites with broken links more strictly.
  • The CSS usage test now considers CSS properties case-insensitive.
  • The CSS usage test now provides a complete exportable breakdown of issues where the CSS does not follow best (ideal) practice.
  • The CSS usage test no longer applies a fractional penalty for using the !important declaration.
  • The CSS usage test now assigns 10% of the overall score based upon the use of external (versus internal) stylesheets. External stylesheets are faster and widely considered best practice.
  • The CSS usage test has been renamed the Stylesheets test.
  • The Design list test is now slightly smarter at identifying different designs. Similarity thresholds are not displayed in results accordingly.
  • The Design list test has been renamed to the Design variations test.
  • The Design variations test now excludes pages which appear to be redirections.
  • The Domain name test is now scored (squatters reduce the score from 10). However, the resulting score does not impact the Summary scores for the website in any way.
  • The list of target bandwidths used by the File usage test has been updated, with the addition of 2MB, 4MB, 8MB and 16MB internet broadband speeds. The default speed is now 1MB.
  • The File sizes test is now called the Speed test.
  • The File sizes test now itemises each page (how large / fast it is to load) as well as individual file sizes.
  • The File sizes test now identifies images embedded using CSS.
  • The File sizes test now identifies missing files individually.
  • The File sizes test no longer displays information about missing/incorrect MIME types, as this information is largely irrelevant for the modern web.
  • The Fonts test now no longer penalises for the use of non-web friendly fonts, but it does highlight their existence. The test now also recognises a much wider array of new fonts, including those specific to the Mac.
  • The Fonts test now recognises style changes in borders for links, background positions (for CSS sprites) and the toggling of visibility (e.g. for some CSS based drop-down menus).
  • The Fonts test now analyses the CSS of links on a per-link basis (previously, a per-page basis). This allows for much more fine-grained distinctions of good/bad link styling.
  • The Fonts test is now the Link states test. Superfluous font-specific scoring has been removed, leaving the core link-state analysis exclusively. Scoring has been completely changed as a result.
  • The Flash usage test now includes a small weighting (10%) for the absence of Flash detection, even when Flash is embedded in a rich HTML page.
  • The Incoming links test has had Bing support added (now removed, due to lack of access to this information in Bing).
  • The Headings test now correctly detects Headings specified by using images with alt text inside the Heading.
  • The Headings test now ignores pages which are redirections.
  • The Links now correctly detects text specified by using images with alt text inside the link.
  • The Outgoing links test now reports the % of pages which link to a given page / domain, instead of the total number of links over the number of pages (which was needlessly confusing). This test does not score, so scores are unaffected.
  • The Popups test has been removed, as it is no longer possible to reliably detect the few remaining types of popups that do still exist (they generally use obfuscation methods to evade popup blockers) and they are much less commonplace, due to the widespread blocking of popups by all major browsers.
  • The Popularity rank test is now the Popularity test. It now checks the popularity rank and the rate of change.
  • The Printability test now excludes pages which look like client-side redirections.
  • The Readability test is now much easier to understand (the irony). The Gunning Fog index has been translated into "Reading Age", which is the same number + 5, and makes more intuitive sense. Other refinements to the results have been made.
  • The Readability test now scores based on the Gunning Fog Index, instead of the FK as before.
  • The Search Score has become Marketing Score, which is expanded to cover Internet Marketing in general (previously, the focus was on SEO, which is now a part of effective IM overall).
  • Metadata test is now non-scoring (the relevant scored aspects are now in the Search engine results test).
  • The Redirections test now detects the window.location.replace() method for client side redirections.
  • The Site structure test now describes the path structure used by the site, listing the number of pages inside each path.
  • The Spelling test now lists the language(s) found within the site, and whether they were tested.
  • The Spelling test now handles some less common variations on language codes, ensuring more pages are tested.
  • The Spelling test now automatically excludes 'learned' brand names and other non-dictionary words from a language corpus of over 1 million webpages.
  • The Spelling test now understands some forms of basic stemming (for example, it will assume "Silktide's" is correct, having learnt "Silktide").
  • The Search engine placement test now shows the number of results from each search engine, where supported.
  • The Search engine placement test now recommends extremely competitive keywords are potentially dropped.
  • The URL format test has been completely redesigned, and now uses more sophisticated and up-to-date criteria for evaluating individual web addresses.
  • The URL format test now displays individual web addresses that are affected, and scores them separately.
  • The Use of keywords test no longer uses the site name of a website as a keyword by default, due to the high level of erroneous data that is typically provided this way. Instead, if no keywords are provided or found, this test will not return a score, and explain the situation.
  • The Use of keywords test now only checks explicitly defined keywords, and ignores those specified in the meta keywords.
  • The Use of keywords test is now smarter at excluding low-value phrases, using a corpus of 150,000 most common English words to determine generic phrasing.
  • The Use of keywords test now displays issues with generic keyword selections, like "software", which are almost ubobtainable.
  • The Use of keywords test now displays the level of competition for desired keywords on Google.
  • The Use of keywords test results have been simplified, with a focus on the highest value information.
  • The W3C compliance test now only displays a summary of what pages have errors, not the individual errors themselves. This data proved to be unnecessary and distracting. As a result this test now uses significantly less space.
  • The W3C compliance test now weights the score based on the relative importance of each page.
  • Many tests now weigh scores based on the relative importance of each page. For example, a usability issue on the homepage is much more significant than the same issue on a deep legal page. Accordingly scores for most tests will change.
  • The count of internal links now accounts for redirections which were previously ignored (affects many tests).
  • All tests are now better at parsing obscure Unicode and UTF-x characters.

Upgrade notes

  • Categories will be removed. You should re-configure sites afterwards manually.
  • Default config will not be updated if you have existing sites, you may wish to review settings (such as target bandwidth) manually.
  • The list of keywords for each site will no longer include [%siteName%], meaning most sites will be blank. You should specify appropriate keywords for each site manually.
Retrieved from "http://siteray.silktidewiki.com/SiteRay_4.0"
Personal tools