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

[Bruce Blackshaw <[e-mail]>]

It's true that you don't always know the size at the start.

Nonetheless, it could be be specified that iff the size of the raw data 
is known, then "size" could be used. In the many cases that this is 
possible, performance should be faster (although I can't quantify by how 
much).

regards

Bruce Blackshaw

On 28/01/2013 4:32 AM, Alun Jones wrote:
>> Of Ángel González
>>
>> Bruce Blackshaw wrote:
>>> Re LOCK, I think it would be useful to have an alternative to
>>> "boundary" called "size", that could provide the size of the raw data
>>> for binary transfers.  That might provide quite a performance
>>> advantage compared to having to search for the boundary.
>>>
>>> regards
>> Good point. I think it would be much more useful than the boundary
>> method.
>> The value of such size parameter should be the amount of data sent on the
>> wire.
> This has an obvious problem - what if the size of the file changes during
> the transfer?
>
> There are many uses of FTP (webcam being the one that comes immediately to
> mind) where a "file" is retrieved, and is handled as if it is a
> potentially-infinite stream. We have a webcam at work that encodes video as
> an animated GIF that never ends. Doesn't display in Internet Explorer,
> because IE is paranoid, but it's a use-case that should be considered. The
> boundary method works for this use, where the size method doesn't.
>
> Alun.
> ~~~~
>
>