Utilize nsIDocumentEncoder for Translations Language Identification
Categories
(Firefox :: Translations, task)
Tracking
()
| Tracking | Status | |
|---|---|---|
| firefox146 | --- | fixed |
People
(Reporter: nordzilla, Assigned: nordzilla)
Details
Attachments
(1 file, 1 obsolete file)
Description
In Bug 1967758 I transitioned translations language identification to use the DOMExtractor code.
Initial profiling is showing that the nsIDocumentEncoder is still faster at extracting text than the DOMExtractor code.
DOMExtractor (2048 code-unit limit)
nsIDocumentEncoder (2048 code-unit limit)
In the short term, I would like to move Translations back to using nsIDocumentEncoder until such a time that we improve the performance and capabilities of DOMExtractor.
| Assignee | ||
Updated•7 months ago
|
| Assignee | ||
Comment 1•7 months ago
|
||
This patch updates the Translations language-detection algorithm
to utilize nsIDocumentEncoder for the page-text extraction.
| Assignee | ||
Comment 2•7 months ago
|
||
This patch adds a new Translations telemetry event
for the language identification, such that we can
get a sense of the average performance of our language
detection code in aggregate.
Updated•7 months ago
|
Updated•7 months ago
|
Comment 3•7 months ago
|
||
Comment on attachment 9521484 [details]
Bug 1995634 - Add translations langId telemetry r=#translations-reviewers!
Revision D269461 was moved to bug 1861698. Setting attachment 9521484 [details] to obsolete.
Comment 5•7 months ago
|
||
Backed out for causing tr8ns failures on browser_translations_full_page_telemetry_auto_translate.js
Updated•6 months ago
|
Description
•