content_encoding: Skip checking support for "none" encoding

[MoSal]

Some servers return a "content-encoding" header with a "none" value.

libcurl does not recognise "none" as a special value, which can cause
unrecognised content encoding type errors.

This patch fixes that issue.

[jay]

https://www.iana.org/assignments/http-parameters/http-parameters.xhtml#content-coding

What servers return Content-Encoding: none? It appears to be non-standard. We could add it as an alias for identity I guess.

diff --git a/lib/content_encoding.c b/lib/content_encoding.c
index 46bef0c..2b2188b 100644
--- a/lib/content_encoding.c
+++ b/lib/content_encoding.c
@@ -726,7 +726,7 @@ static void identity_close_writer(struct connectdata *conn,

 static const content_encoding identity_encoding = {
   "identity",
-  NULL,
+  "none",
   identity_init_writer,
   identity_unencode_write,
   identity_close_writer,

[MoSal]

What servers return Content-Encoding: none?

Forums using ElkArte is the use-case I found in the wild.

For example:

curl --compressed -O  'https://www.elkarte.net/community/index.php?action=dlattach;topic=4652.0;attach=4971'

[bagder]

I googled for "Content-Encoding: none" and that was a sad experience. Lots of hits since many years back.

It seems this header (value) has been proposed and advocated as a solution to a problem where "identity" was recognized but got the wrong treatment. Usually for PHP solutions, which the elkarte.net server mentioned above also uses (running on nginx).

So, I'm mildly positive to recognizing this as an identity alias.

[mkauf]

I have found this: getgrav/grav#548 (comment)

So Content-Encoding: none is a hack to prevent Apache's mod_deflate from compressing the response. I don't know how widespread this hack is.

[jay]

Ok just landed. Thanks guys