User-agent sniffing tends to break on future browser versions or browsers you don't know about; should use feature detection instead.
Agent sniffing can be sometimes appropriate to blacklist particular things we know are bad, but should be much more limited and explicit.
Version: unspecified
Severity: normal
| Title | Reference | Author | Source Branch | Dest Branch | |
|---|---|---|---|---|---|
| Include docroot/noc in the image we build | repos/releng/release!38 | oblivian | add_noc_to_images | main |
(This seems to be mostly used in checking whether we need to save/restore the
scroll & selection state. Since an IE-specific interface is being used there,
it may make more sense to base the check on that.)
This is being used for a variety of things, which can not be feature detected, including CSS layout issues.
Yes, feature detection is superior, but it's not always possible.