IEContentAnalyzer was introduced to deal with content sniffing of resources on our potentially arbitrary user uploads.
IE8 introduced X-Content-Type-Options: nosniff, now supported by the major browsers, which if we would use it on all resources (esp. also uploads.wikimedia.org) should remove the need for IEContentAnalyzer if I understand it correctly.
Browsers listen and obey this header for all uris when they are attempted to be loaded as script or style resources. In addition the header triggers CORB protections if supported by the browser.
- https://fetch.spec.whatwg.org/#x-content-type-options-header
- https://chromium.googlesource.com/chromium/src/+/master/services/network/cross_origin_read_blocking_explainer.md#Observable-CORB-impact-on-images
MediaWiki already returns x-content-type-options:nosniff` for all requests, but upload.wikimedia.org does not return results with X-Content-Type-Options: nosniff.
Removing IEContentAnalyzer was one of the notes of T232563: Drop IE6 and IE7 basic compatibility and security support. It is known that IEContentAnalyzer creates problems, it is the cause of T182264: IE content analyzer is executed on non-first chunks during chunk uploading, preventing legitimate uploads, which might actually explain some of the many user reports about failures with using chunked uploading. It is also blocking T28059: Add support for KML/KMZ filetype.