From: "Ian Clarke" <[EMAIL PROTECTED]>
Thanks for your questions, Ian.
Please take the time to digest my response.
>I have some concerns over this approach. What is required for someone to
>access FreeWeb? I get the impression that it:
>* Is Windows only
I've had reports of it working in Linux under Wine. But that's not an
excuse.
I have only recently learned some of the basics of Windows GUI programming,
and know nothing about native Linux GUI. To run there, much or most of it
will need to be re-written for (say) KDE or Gnome. You'll note from the
website (http://freeweb.sourceforge.net) that I'm inviting experienced KDE
and Mac programmers. If no-one steps forward, then I'll go ahead in time and
learn KDE GUI programming myself and cut a Linux version. As for a Mac
version - that'll take longer, since I don't have a Mac and am hard put to
justify purchasing one.
There is absolutely *NO* intention of it remaining Windows-only.
Windows is just a convenient platform for the initial proof-of-concept,
which this alpha release essentially is.
>* Is closed source (I find the argument that the code is too disorgainised
> to release yet to be somewhat unconvincing)
FreeWeb uses JNI to invoke Freenet java methods, particularly CLI clients,
and freenetmirror.
Note that it sets the "freenet.CLI.library" flag, which has been removed in
the latest release of Freenet - without this flag, FreeWeb will fall over in
the middle of publishing sites. (I did ask on dev for the library flag to be
left intact for a while - this was ignored) Thus, FreeWeb requires a
retrograde Freenet - 0.3.7.0 thru 0.3.8.1, or else it won't work. :((
Currently I'm developing an FCP C library called ezFCPlib, a portable,
programmer-friendly API, which will replace the whole back end. Expected
time of implementation (with support of much of the old and new freenet
metadata) is one week. On the surfing side, FreeWeb uses FProxy - I'm also
to replace the proxy's back end with calls to ezFCPlib, and am rewriting 80%
of the proxy server code (in response to complaints of the proxy code
blocking cookies, plus my total dissatisfaction at the present state of the
proxy code).
I could release source now, but it won't work with latest freenet, also, it
will have little architectural resemblance to what FreeWeb is very shortly
to become. The source as it stands will be misleading. Plus, it will
encourage bad freenet programming practice.
But if people find the delay in source release unacceptable, even given the
above, I'll release it. Be warned though - it's a mess. I'd probably put it
out in a 'daily zip' format. I would much rather put it out when I've done
the migration to FCP. In medium and longer term, what's the problem with a
slight initial delay. One year from now, who's gonna give a damn that I took
3 or 4 weeks to release source?
>* Relies on some central DNS server to convert .free domains to FProxy
> domains
That is *NOT* true!
FreeWeb includes a program called 'fwgetdns', which reads all the keys in
the FreeWeb in-freenet keyindex and converts them into DNS records for a
user's local cache. The doco explicitly tells the user how to delete this
cache, in situations where privacy of access to their PC may be compromised.
Therefore, this 'some central DNS server' is a convenience, not a
requirement.
When a site is published, a DNS record with the site's SSK public key and
the site's domain name is written to this key index. A 'DNS daemon' harvests
these records and writes them to another SSK - the FreeWeb DNS SSK, whose
public key is hard-wired into the FreeWeb software. I trust that I'm not
going to be asked to explain why the FreeWeb DNS SSK's private key is not
also built into FreeWeb ;)
I am in total agreement with Brandon's supportive correspondence, and will
be adding a facility whereby anyone can set themself up as a DNS registry.
In this scenario, FreeWeb DNS registrars would be encouraged to enable their
registries to harvest all records from all other registries, so that every
registry ideally has all records. Also, end surfers can nominate a 'hit
order' for registries - for all .free requests, they can nominate the search
order among the available registries. If one registry starts suffering DOS
attacks, say, users can drop that registry from their list.
Now that's another topic - FreeWeb DNS is totally vulnerable to DOS attack
if the name of the key index is easily known. But the only alternative is to
use some out-of-band means for DNS registration.
>* Mandates that anyone who wishes to use a .free domain uses FreeWeb, thus
> forcing them to tolerate the above-mentioned restrictions.
I can see how one could form such a perception. That is certainly *not* the
intention.
To allay fear and loathing, I'll put up an algorithm document which will
allow anyone to access FreeWeb.
But you don't even need this.
To publish sites to .free, and to surf such sites, all anyone needs to know
is:
1) the in-freenet key index 'freeweb0.1a-1 contains the DNS records of all
published FreeWeb sites. These records are in the form:
freeweb:sskpublickey/domainname:sitedescription
Do a list on this index now, and see what I mean.
2) *Any* freenet site visible as MSK@SSK@sskpubkey/subkeyname// is
compatible with FreeWeb, once it is listed on the FreeWeb key index, which
anyone can do (and which I recently offered to do on anyone's behalf)
3) *All* FreeWeb-published sites are accessible via mainstream freenet URIs.
Within FreeWeb, invoking the 'properties' menu item on any site (or
double-clicking on that site) shows a dialog with, amongst other things, a
mainstream freenet URI under which the site will be visible. Alternatively,
looking in the fwpublish.ini file reveals all.
>FreeWeb seems to add a very thin layer around Freenet, and freenetmirror,
>but in the process adds a number of rather undesirable requirements.
>Can you clarify this David?
On the other hand, I'm getting criticised by windows users for the alleged
'technical complexity' of FreeWeb.
An old school friend, who is now a windows app development manager and
exquisite hacker, found aspects of FreeWeb hard to understand. He suggested
that I bundle Freenet in with the one install, and not require freenet to be
installed separately. As someone who was writing drivers in assembler at the
age of 14, I don't see him as technically lacking.
Yes - FreeWeb *is* a very thin layer. That's not news.
FreeWeb's core concept is the use of an actual http proxy which creates a
realistic simulation of a TLD called .free - It sources freesite pubkeys and
subkeys to make these freesites visible as if there actually *is* a ".free"
TLD, but forwards http requests on all other domains out to the mainstream
web.
Rather than viewing FreeWeb with what seems to be suspicion and prejudice
(forgive me if I'm wrong here), one might like to take note of the interest
it is generating in Freenet. I saw a /. post saying something along the
lines of "I'm a programmer, but I'd perceived that Freenet wasn't properly
up and running. FreeWeb has inspired me to take a second look at Freenet."
(not an exact quote).
If one looks inside inform.php, one might notice a growth in the number of
running freenet nodes.
It is part of my "agenda" (!) that FreeWeb have a net supportive effect on
Freenet.
But what fascinates me from these questions is the huge gulf which exists
between the linux hacker community and the windows user community. It's like
judaism versus islam.
In the perfect world, everyone would immediately throw Windows in the bin,
learn to program, and immediately switch to Linux.
In the perfect world, all school curriculums from infancy would teach
programming, security, encryption, etc.
In the perfect world, any grandmother would write new drivers for a scanner
that's playing up, and be adding stitches to the knitting while she's
waiting for the code to compile. She'd also be writing VR apps for her
grandchildren while she's waiting for the scones to bake.
Could take a while to get there.
There *is* a demand for fast access without a huge up-front learning curve;
for example, with cars, pour in gas, sometimes change oil and water, and
you're generally OK. FreeWeb is intended to cater to this demand, while
still encouraging people to visit the Freenet site and learn about pure
freenet.
Also, even excellent attempts at user-friendly linux (like Mandrake 7.2)
still fall short of the ease of Windows.
Compare the required steps for adding new fonts, for example.
In conclusion, many of the criticisms boil down to the fact that I'm not
making things happen fast enough. There is a definite evolution plan for
FreeWeb which will annihilate all these concerns. Sorry if I'm not bringing
it all in yesterday. Hey, I'm stealing hours through the day, 7 days a week,
to learn, document and program as fast as I can. As it is, I get frequent
strange looks from my wife at the volume of time I'm spending at the PC.
If you or anyone else need further clarification on any of these areas,
please don't hesitate to ask.
Cheers
David
_______________________________________________
Chat mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/chat
