Re: [ftpext] New Version of FTP HASH, RANG, LOCK, and HOST

[Anthony Bryan <[e-mail]>]

Daniel Stenberg wrote elsewhere:

"I agree that sending data in a chunked-encoding style (from HTTP 1.1)
is much easier for both sender and receiver instead of having to check
for a boundary separator in the stream.

It would be good to get an FTP server author's point of view on this."

any other +1s or opinions?

On Sat, Feb 2, 2013 at 12:59 AM, Hans Andersen <[e-mail]> wrote:
> I just wanted to back my colleague, Bruce, on the performance issue related
> to detecting boundaries.  A few weeks ago I wrote code that detects MIME
> boundaries in HTTP-POSTs while piping the data to a file.  The best I
> managed to achieve was significantly slower than simple piping.  I therefore
> think that optionally including the size is important for performance.
>
> On 2 February 2013 14:26, Anthony Bryan <[e-mail]> wrote:
>>
>> right, we could leave the boundary REQUIRED but have an OPTIONAL size
>> that would be used (probably the majority of the time?)

--
(( Anthony Bryan ... Metalink [ http://www.metalinker.org ]
  )) Easier, More Reliable, Self Healing Downloads