Easyling release notes November, 2023

November came and went with only a minor new feature and a bunch of fixes. The new feature is rather exciting though: When you use the Crawl wizard, it’ll check the user agent of your browser and set it as the default value for Custom user agent under the Miscellaneous tab of step 4. This will allow you to crawl more sites. Read the full details after the fold.

Custom user agent

There is a little life hack that you may not know about. In the Crawl wizard, on step 4, there’s a Custom user agent field. This allows you to specify the user agent that the crawl should use. For example, if you want the crawler to pretend to be the latest version of Firefox on Windows at the time of writing, you can just enter Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:120.0) Gecko/20100101 Firefox/120.0. When you do that, the server of the original website will see the following user agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:120.0) Gecko/20100101 Firefox/120.0 AppEngine-Google; (+http://code.google.com/appengine; appid: u~skawa-easyling).

This can be beneficial for crawling websites that check the browser version and adapt the content appropriately. If you leave this field empty, Easyling will fall back to an older version of Google Chrome. With the update we added this month, this field will be pre-filled with your user agent, likely a new version of Chrome of Firefox, ensuring that the right content is picked up.

Note that AppEngine-Google; (+http://code.google.com/appengine; appid: u~skawa-easyling) will always be added to the end due to the underlying Google infrastructure.

New defaults for JavaScript exports

When creating a JavaScript export for Crest, the default is now to export entries with translations as opposed to all entries.

This reflects our change in philosophy. When Client-side translation, the older version of Crest, was first envisioned, the plan was to work as follows:

1. Create a JavaScript-type export that contains all translations.
2. Load said file in the web browser.
3. Apply these translations.

However, if configured by a member of our support team, Crest can do an additional step: When translations for a segment aren’t included in the JS export, Crest can ask the Easyling servers whether translations are available.

This is great for projects needing constant updates. This way, the system can find new content, pipe it through Auto pre-translate and apply the new translations automatically. However, if a segment is included in the JS export, the translation there will be considered to be the only one available. It won’t query the servers for updated translations. The result of this is that if all entries are included, there may be segments that do not yet have translations. When these translations become available, they will not show up until a new export is created.

With the new default, the entries without translations won’t be included, so the system will check for updated translations every time ensuring that such translations show up as soon as they become available.

If you wish to try the new “missing fetch” feature, please drop us a line. Otherwise, the old behaviour is still available.

Fixes

This is where our main focus was this month, but they are reasonably small, so I’ll try to just list them off here:

• New statistics are now displayed on the Request summary card in the Crawl list: Pages that couldn’t be reached at all, not even with a simple HEAD request.
• We improved the Create new project dialog so that it no longer rejects a few valid domains by saying they were invalid.
• During TMX import, when a translation is too big to save into the Translation Memory, the entry will now be skipped instead of stopping the whole population process.
• We fixed a bug that prevented XTM settings from loading as expected when no previous settings were saved.