This is a decade-old bug that I've just never taken the time to properly report before now. Try to load https://bug-254326-attachments.webkit.org/attachment.cgi?id=465550&action=review in WebKitGTK and the load will fail with error message "Load cannot follow more than 20 redirections." If we try the same thing with curl: $ curl -L 'https://bug-254326-attachments.webkit.org/attachment.cgi?id=465550&action=review' curl: (47) Maximum (50) redirects followed So WebKitGTK has the same behavior as curl. Problem is https://bug-254326-attachments.webkit.org/attachment.cgi?id=465550&action=review redirects to https://bugs.webkit.org/attachment.cgi?id=465550&=&action=review, but that itself redirects back to https://bug-254326-attachments.webkit.org/attachment.cgi?id=465550&action=review, so there's no winning.
This happens when the attachment isn't a patch. It would be nice to make the redirection make more sense indeed.
<rdar://problem/107421556>
How does one gets to this link that includes action=review?
Bugzilla emails contain links with "action=review" for every added attachment.
(In reply to Alexey Proskuryakov from comment #4) > Bugzilla emails contain links with "action=review" for every added > attachment. Proposal: we could simply stop adding this to the links in emails? There are very few patches being landed via Bugzilla nowadays, so the desired action= is almost never going to be "review". Then most people would never notice this bug.
We can probably make the appending conditional almost as easily. The review functionality is all ours, not coming from Mozilla, so whoever implemented this must have made too big of a simplification.