dcdev mailing list

Direct Connect developers, 2003 to 2005
← All threads

Question

4 messages · [email protected], Todd Pederzani, eric
23 January 2004, 01:17[email protected]
Why is it we are dismising an xml protocol? it is the obvious choice it sounds like since we want the protocol to be extendable, and yes i have read the arguments in the past but they make no sence.
The main reson xml was dismissed was the "bandwidth" that seems like a logicle argument, until you choose textbased protocol over binary and start adding 2 byte \r\n instead of 1. as far as i can tell the current ADC protocol dosnt seem that great and i think xml would be much better for what we want.

But I really dont care what protocol is chosen I only have a couple requests/suggestions
 
 In Search Results Send an extra integer defining clients inet speed so the info dosnt need to be  retrieved from the server userlist(although this result isnt necicarly true at least it will save some
 cpu while parsing).

 Also in search results I think a Feild should be set aside for fileinfo  ie movie dimentions or music  bitrate and time.

--------------------------------------------------------------------------------
----------------------------------------------------
On a side note. If somone has the ability would you please make a dll that i can call from VB6 for TTH hashing I currently dont have the time to figure out a native implementation into VB.

Timothy Marin , IntraDream.com
23 January 2004, 02:01Todd Pederzani <[email protected]>
to Direct Connect developers

[email protected] wrote:

Also in search results I think a Feild should be set aside for fileinfo  ie movie dimentions or music bitrate and time.

Well, here's one way to format metadata on search results in xml:

http://rfc-gnutella.sourceforge.net/developer/testing/metadata.htm

On a side note. If somone has the ability would you please make a dll that i can call from VB6 for TTH hashing I currently dont have the time to figure out a native implementation into VB.

Here's a non-free library:

http://www.slavasoft.com/quickhash/

I did about 15 minutes of googling, but that's the best I can come up with at the moment.

Share & Enjoy
- Todd

23 January 2004, 04:55eric <[email protected]>
to Direct Connect developers <[email protected]>, [email protected]

But I really dont care what protocol is chosen I only have a couple
requests/suggestions

  In Search Results Send an extra integer defining clients inet speed so
the info dosnt need to be
  retrieved from the server userlist(although this result isnt necicarly
true at least it will save some
  cpu while parsing).

Perhaps we should take the problem from a different point of view: is a user list really necessary ? Most of the time, we will "private-chat" with either someone we have received a search reply from or someone we already knows. A command like "$IsOnline nickname[,nickname...]|" is probably better (moreover, the hub can notify a client when the user with the given nickname enters).

  Also in search results I think a Feild should be set aside for fileinfo ie movie dimentions or music
  bitrate and time.

It looks good but these informations are meaningless (quality of non-lossless format mainly depends on the quality of the initial source), fairly hard to obtain (you must process all shared files) and will create clients which are unable to be maintained because there is dozens of file type and new ones are regularly available.

Eric

24 January 2004, 03:08[email protected]
to Direct Connect developers
Sorry Eric For The Private Responce**

Quoting eric <[email protected]>:


> But I really dont care what protocol is chosen I only have a couple
> requests/suggestions
>
>   In Search Results Send an extra integer defining clients inet speed so
> the info dosnt need to be
>   retrieved from the server userlist(although this result isnt necicarly
> true at least it will save some
>   cpu while parsing).

Perhaps we should take the problem from a different point of view: is a user

list really necessary ? Most of the time, we will "private-chat" with either

someone we have received a search reply from or someone we already knows. A command like "$IsOnline nickname[,nickname...]|" is probably better (moreover, the hub can notify a client when the user with the given nickname

enters).


Hmm i think you may have misunderstood the request i agree a filelist isnt really that importatant my point is as far as i know in a "$SR" no inet speed is sent so you can only fill the "Speed" column by checking against the server userlist for each result.

>   Also in search results I think a Feild should be set aside for fileinfo > ie movie dimentions or music
>   bitrate and time.

It looks good but these informations are meaningless (quality of non-lossless

format mainly depends on the quality of the initial source), fairly hard to obtain (you must process all shared files) and will create clients which are

unable to be maintained because there is dozens of file type and new ones are

regularly available.


i wouldnt say meaningless, mp3's for example i feel is safe to say likely the initial source is a cd so bitrate would be nice to know and track time is cool no matter the quality for movies or music.. and dimentions although not a valid guage for quality may help. also i beleave the codec refrance is in an avi file header might be nice when choosing what version of a movie you want.
when i said "a Feild should be set aside for fileinfo"

i think it should just be one field so there can be either no info or just helpfull info about the file if available.. i would prolly only support certin types in my client if it where available because im not up for writing a ton of header info functions.. i have writen avi/mp3 already.
(i dont think we need result|bitrate|freq|dimentions|time  added in the protocol result|fileinfo[|Speed ;o)] i think would be good.. maybee delimit fileinfo with a different chr or not just read it as is and maybee show it as a tooltip in results listview.)

on another point(on a napster client i was working on and am currently updating) loading of 3500 mp3's with header info takes 15-20 seconds on my computer. without tag info <2 seconds. so i wrote a caching function that saves a basic txt file with the info .. loading header cache + 3500 mp3's takes <3 seconds(cache also will detect and get new info for uncached headers).

another thing we know is users are stupid they like to see file info it makes them feel like they know what there doing. :o)

Eric



Tim-