Re: freenet performance

Philippe Landau wrote:
> Alex R. Mosteo wrote:
> 
>> Alex R. Mosteo wrote:
>>
>>> Caco Patane wrote:
>>>
>>>>>> However the current 0.5 works pretty well IME.
>>>>>
>>>>> that's great news.
>>>>> how is performance compared to for example aMule or eMule ?
>>>>
>>>> Freenet is slow (my experience), is nice for text-based stuff
>>>> (Freesites) and a few images. I have no experience retriving files
>>>> with more than 5MB of size.
>>>
>>> My experience has been fairly good. Using Frost, I'd say I've 
>>> retrieved  > 90% of the large files I've tried. And the speed is 
>>> usually quite good, again IME. Specially, for people used to wait in 
>>> queues in eMule it shouldn't be a new pain. The Frost progress bar is 
>>> mesmerizing to me...
>>
>> Sorry, where it says Frost it should say Fuqid.
> 
> so what would be your recipe for installing such a freenet ?
> i am planning to repost a step by step instruction if you don't mind.

I don't think I'm an authoritative source of information about this. All 
I did is:

1. Install latest stable freenet.
2. Have it running all day if possible.
3. Install fuqid for convenience (i.e you can download without it 
anyways, it just simplifies retries and such). I think it also allow 
insertions, but I have never used it for that.
4. You may want to use FIW for your site insertions.

>>> I guess the real pain is in inserting large files. I used to insert 
>>> some text-only small sites and the time required wasn't negligible.
>>> In truth and IMO, a popular release site a-la somethingreactor is 
>>> something bound to happen sooner or later, here or over I2P or 
>>> something else.
> 
> that would be a server announcing new inserts ?

If I had to construct such a portal, I would have an edition site 
publishing new links. People could submit these with NIMs (a freenet 
basic messaging system), I suppose. This obviously requires some manual 
work or else coding.

A drawback is that you can't have search capabilities within freenet, so 
I'd publish in that site periodically a full database of 
links/descriptions, and maybe bundle an open source search tool or 
something. If this database were to grown too big some alternative could 
be studied, like sharing link references in some non-anonymous, 
searchable P2P network.

I think I'm not saying anything new here. I guess this has no appeal to 
current communities because there's no way to collect info about users 
to shell to 3rd parties, or to place tracked ads unless you have a part 
of the site pointing to externals, so losing part of the anonymity. Also 
forget about scripting of any kind. It should be plain HTML.

Another thing is that maybe the current freenet is not prepared to 
support the huge amounts of data that would have to be inserted. Some 
statistics about the usual datastore sizes could be useful here.

_______________________________________________
chat mailing list
[email protected]
Archived: http://news.gmane.org/gmane.network.freenet.general
Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/chat
Or mailto:[email protected]?subject=unsubscribe