Hi Chandru, I think there will be occasions -- e.g. replicating compressed
attachments -- when we won't want to unzip the response body. We are
decompressing other gzipped responses using a little wrapper much like what you
pasted below. I suppose if ibrowse handled this directly it would save a few
lines on our end. All the best,
Adam
On Mar 2, 2010, at 6:39 PM, Chandru wrote:
> Hello everyone,
>
> I'm the author and maintainer of ibrowse. I know you use ibrowse in CouchDB
> and I had an idea a couple of days ago.
>
> Would it help improve performance if ibrowse supported receiving compressed
> data from the webserver but made it transparent to the calling process? It
> is simple enough for the caller to do (as shown below) it but would it help
> if ibrowse supported that natively?
>
> get_url(Url) ->
> case ibrowse:send_req(Url, [{"accept-encoding", "gzip"}], get,
> [], [{response_format, binary}]) of
> {ok, "200", Headers, Body} ->
> Headers_lc = [{string:to_lower(X), string:to_lower(Y)} || {X, Y}
> <- Headers],
> Body_1 = case proplists:get_value("content-encoding", Headers_lc)
> of
> "gzip" ->
> zlib:gunzip(Body);
> _ ->
> Body
> end,
> {ok, "200", Headers, Body_1};
> Res ->
> Res
> end.
>
> cheers
> Chandru
