Re: Difference between LWP::UserAgent and AnyEvent::HTTP

After doing packet capture, I determined that AnyEvent::HTTP is sending the header 'X-AcceptDeltaUpdates' as 'X-acceptdeltaupdates', which the server ignores.

LWP useragent retains the case you feed it.

I believe in general the problem is that AnyEvent::HTTP should not be applying the line $hdr{lc $k} = $v to headers that start with "X-"? Anyone know HTTP spec says about this?

Comment on Re: Difference between LWP::UserAgent and AnyEvent::HTTP Download Code

Replies are listed 'Best First'.
Re^2: Difference between LWP::UserAgent and AnyEvent::HTTP by haukex (Archbishop) on Sep 06, 2021 at 04:37 UTC
The HTTP/1.0 and 1.1 RFCs both state that header names are case-insensitive. I can't find any information on `X-AcceptDeltaUpdates`, but IMHO the server should probably accept the lowercase version as well.	[reply] [d/l]
Re^3: Difference between LWP::UserAgent and AnyEvent::HTTP by afoken (Chancellor) on Sep 06, 2021 at 17:53 UTC
I can't find any information on X-AcceptDeltaUpdates IIRC, MIME (used both for Mail and for HTTP) allows to invent non-standard "experimental" headers by prefixing them with "X-". The other end should silently ignore all unknown experimental headers. Alexander -- Today I will gladly share my knowledge and experience, for there are no sweeter words than "I told you so". ;-)	[reply]
Re^4: Difference between LWP::UserAgent and AnyEvent::HTTP by haukex (Archbishop) on Sep 06, 2021 at 18:58 UTC
IIRC, MIME (used both for Mail and for HTTP) allows to invent non-standard "experimental" headers by prefixing them with "X-". Correct, though its use is now discouraged - from RFC7231 (emphasis mine): Authors of specifications defining new fields are advised to keep the name as short as practical and not to prefix the name with "X-" unless the header field will never be used on the Internet. (The "X-" prefix idiom has been extensively misused in practice; it was intended to only be used as a mechanism for avoiding name collisions inside proprietary software or intranet processing, since the prefix would ensure that private names never collide with a newly registered Internet name; see [BCP178] for further information). And the linked RFC6648 is dedicated to discussing the deprecation, with some interesting background in the appendicies.	[reply]
Re^3: Difference between LWP::UserAgent and AnyEvent::HTTP by sectokia (Friar) on Sep 06, 2021 at 12:07 UTC
Yes it appears HTTP servers should do case insensitive matches to headers by specification. However it sort of appears by convention that HTTP libraries should not change case of headers that are specified to them. Lesson learned: Never trust either end!	[reply]