Skip to content
This repository has been archived by the owner on Jan 10, 2023. It is now read-only.

Latest commit

 

History

History
132 lines (82 loc) · 7.43 KB

CHANGELOG.md

File metadata and controls

132 lines (82 loc) · 7.43 KB
Raw

Changelog

v0.5.0 (23 October 2018)

Changed

  • The engine's config (sirko.conf) was migrated to a Toml format. Settings didn't change their position in the config file, so it should be easy to find and specify your settings. Don't forget to backup your current config before upgrading.

Added

  • If you use Google Analytics (GA) on your site, now it is possible to import sessions from GA, thus, you can get more realistic predictions earlier. Please, get more details here (sirko-io/engine#40).
  • Besides prefetching JavaScript and CSS files, the client was changed to prefetch images too. However, only critical images makes sense to be prefetched. So, a CSS selector for finding images should be provided while setting up the client. (sirko-io/client#36).

v0.4.1 (23 April 2018)

Fixed

  • The client only sends paths of pages to the engine. For example, https://example.org/blog is a URL of the current page, in this case only /blog gets sent. The path is used to precache a page. A bug in code cutting the origin off produced broken paths. For example, if a URL was like this https://example.org/bar/foo/main.html, only /main.html would've been sent to the engine. It leaded to precaching pages which didn't exist. Now the client takes paths as they are (sirko-io/client#33).
  • Some browsers embed internal assets (for example, chrome-extension://bmdblncegkenkacieihfhpjfppoconhi/scripts/in-page-script.js) into pages. URLs of those assets were gathered by the client and later there was an an attempt to precache them. Apparently, they cannot be precached, thus, browsers threw errors. Now the client only gathers URLs which lead to the valid HTTP protocol (sirko-io/client#30).

v0.4.0 (18 March 2018)

Changed

  • Prefetching several pages might be fine for some sites. So, the engine got changed to return a few pages which get prefetched by the client then. A max number of returned pages is controlled by a sirko.engine.max_pages_in_prediction setting. The sirko.engine.confidence_threshold setting is still considered. For example, if the sirko.engine.confidence_threshold is 0.5, only 2 pages can be returned even though the sirko.engine.max_pages_in_prediction is 10.

Added

  • All prefetched resources aren't removed from the cache anymore. Now they are moved to an offline cache which is used when a user is offline. The offline cache accumulates all prefetched resources during the user's navigation. Thus, the site can partially work offline (sirko-io/client#29).

v0.3.0 (1 January 2018)

Fixed

  • Prefetching a wrong page for an authorized user because of missing cookies (sirko-io/client#26)
  • A number of issues related to HTTP requests modifying data (#10, #15, sirko-io/client#1). Now, if there is a request modifying data in a user's transition, the transition won't be tracked, hence, a stale page won't be prefetched.

v0.2.0 (1 October 2017)

Changed

  • The engine doesn't prerender pages anymore. The previous release provided fallback to browsers which don't support the prerender hint. Since the Chrome team deprecated the prerender hint, that fallback turned to a main solution. Now, the client side of Sirko Engine prefetches a predicted page and serves it from the cache once the user visits that page. It means if you already use v0.1.0, the following code can be removed:

    sirko('useFallback', true);

    Also, it is mandatory to serve a sirko_sw.js script from the root of your domain, example:

    https://demo.sirko.io/sirko_sw.js

    Please, get more details here.

  • Performance improvements

Added

  • When a page gets prefetched via a service worker, assets (images, JavaScript, CSS files) aren't fetched until the page gets shown. To speed up page loading, the engine got enhanced to store urls to assets (only JavaScript and CSS files) of a page. So, when the engine predicts a next page, the client prefetches the predicted page and assets of the page. Thus, when the user navigates to that page, the page and assets get served from the browser's cache.

v0.1.0 (15 July 2017)

Changed

  • The engine got migrate to a bolt protocol to work with Neo4j. This protocol provides better performance. The change affects the config file. To specify the url to a Neo4j instance, use the following format in the config file:

    bolt://localhost:7687

    Credentials are a part of the url:

    bolt://username:password@localhost:7687

    The following settings got removed:

    neo4j.basic_auth.username
neo4j.basic_auth.password

    Also, there is a new setting which points either to use a secure connection to the Neo4j instance:

    neo4j.ssl = true

Added

  • Each Neo4j query gets logged when the log level is info. It will help in debugging slow queries.

  • Duration of keeping expired sessions in the DB can be configured via sirko.engine.stale_session_in setting. To get more details about this setting check the config/sirko.conf file.

  • A new sirko.engine.confidence_threshold setting is introduced to define a confidence threshold which must be met to prerender a predicted page. Use this setting to reduce load on your backend if you get some because of the prerendering.

  • The client side got a fallback solution to the prerender hint. That solution prefetches the predicted page in browsers which don't support the prerender hint. It doesn't provide the instance load, for example, in Firefox, but Firefox users get the improved experience anyway. The technical details of the implementation can be found in this article. In order to use the fallback, enable it on the client side:

    sirko('useFallback', true);

    Since it is based on a service worker, the site which uses the engine must be served over the secure connection. Also, the site must serve a service worker script from the root of its domain, example:

    https://demo.sirko.io/sirko_sw.js

    The easiest way is to proxy the request to the engine. If you use Nginx, here is an example:

    location = /sirko_sw.js {
  proxy_pass http://127.0.0.1:4000/assets/sirko_sw.js;
}

Fixed

  • Missing transitions when a user with the expired session continued navigating site (#29).

Removed

  • The starting point which was described in this article got removed. It didn't bring any value to the prediction model.

v0.0.2 (27 March 2017)

Fixed

  • An issue in IE11 which broke the sirko client for IE11 users.
  • An issue with starting the project when the url to a Neo4j instance had a trailing slash.