Low entropy User-Agent config #10
Replies: 5 comments
|
Microsoft Azure powered sites ("bing.com", "vscode.dev" etc.) might block Firefox User-Agent without plausible "Gecko/" value: [ [1] [[ |
|
Twitter expects "Firefox/999" alike. |
|
SiteGround asserts Firefox User-Agent beginning with "Mozilla/"i... |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Up: https://bugzilla.mozilla.org/show_bug.cgi?id=1868415#c2
The `
Accept: ,*/*` is for:[Outdated] [Bug] Repository deep link returned JSON
https://github.com/orgs/community/discussions/70778
.
And [ https://i.redd.it/0czt1ekqww531.png ] alike:
`
Accept: text/html,*/*` coerces HTML page...Requires `
Accept: text/html,image/*,*/*` alike.[ ^
(20250413)
"
image/*" seems no more working.However `
Accept: text/*,image/*,*/*` seems to workaround...(unsure if it breaks other things) ]
GitLab's "crippled" [1] design expects alike `
Accept: ,*/*`...E.g. https://gitlab.torproject.org/tpo/applications/mullvad-browser/-/issues
.
`
Accept: */*` differed the page itself; and glitched scripts handling.(failed assertion on the page's content)
E/> TypeError: URL constructor: undefined is not a valid URL.
("vue.runtime.esm.js"#L3065c16)
@ P ("url_utility.js"#L413)
@ isIssuableUrlExternal ("issuable_item.vue"#L149)
...
[ ^
This seems to break the whole GitLab... Woefully.
(not necessarily the same error message, but alike cause)
Further tested:
https://gitlab.com/dominicp/tutanota-plain-text-bookmarklet
https://gitlab.com/dominicp/tutanota-plain-text-bookmarklet/blob/main/README.md
https://gitlab.com/dominicp/tutanota-plain-text-bookmarklet/blob/main/bookmarklet.js
Might be related:
https://gitlab.com/gitlab-org/gitlab/-/issues/407534
https://gitlab.com/gitlab-org/gitlab/-/issues/345765
https://gitlab.com/gitlab-org/gitlab/-/issues/?sort=updated_desc&state=all&in=TITLE&search=Accept+header
(so broken parameter handling...)
[1]
Pardon the wording. Probably not really so "crippled".
Merely quite irritated by that pointless "/-/" and clumsy search parameter handling.
Plus the garish UI look...
[2]
In fact the response header of GitLab even explicitly stated:
"
Vary: Accept"; indicating which as even intended..?
(but for what really?) ]
[ ↓ Tested: 20240914; no more as of 20250413. ]
[ https://blog.hubspot.com/ ] reported HTTP 400 Bad Request:
When using `
Accept: ,*/*` with browser-like UA.Workaround would be not sending User-Agent.
Or use "
*/*" with browser UA (recommended).[ ^
The primary reason is to reduce uniqueness of the HTTP header:
Making that more like from typical `curl` alike.
(alike automated traffics are actually the primary part of Internet) ]
See also:
[Suggestion] Request headers should not contain more than just-necessary entries
https://forum.vivaldi.net/topic/77666
All reactions