On Friday 18 Feb 2011 19:33:37 Dave U. Random wrote:
> On Tue, 15 Feb 2011, Matthew Toseland <t...@amphibian.dyndns.org> wrote:
> >On Tuesday 15 Feb 2011 19:14:08 Dave U. Random wrote:
> >> Seen on 0.5
> >
> >People still use 0.5? I wonder if anyone has volunteered to provide securit=
> >y support...
> Yes. i estimate there are well over 1,000 nodes with more being added daily

Can you answer the second part? Is the source code being maintained? Has anyone 
competent had a serious look at the HTML filter?
> >> >Toad is still in the same mood.  The network has been screwed for months
> >> >and gets worse each update - currently about 3 per day. =20
> >
> >This is partially true. However there has been significant progress on anno=
> >uncement and UOM towards the end of last week. The UOM may help to bring ba=
> >ck a significant proportion of the lost nodes, which I believe to be old fo=
> >rgotten nodes with update enabled but for which it was failing due to bugs =
> >introduced around the time of the new packet format changes.

Also I have fixed a lot of bugs recently via the testnet. Most of these relate 
to timeouts of various kinds, so should improve performance.
> I've watched this list since 0.7 was forked and the rate of new things
> breaking the network has never ceased to amaze me. How can 0.7 be
> considered even remotely stable? excess of 50 builds since 1 jan 2011 is
> NOT stable by any stretch of imagination. is a sign something is seriously
> flawed from the start.

No, it's not. It's a sign that it's not finished yet. But it's already way past 
0.5 in so many different areas.
> 0.7 should not have been made default freenet download.  should have been a
> separate 'testing' network so new users would not be exposed to the
> continual breakage and multiple updates per week, sometimes multiple
> updates per day.

This was true for a period. However at some point 0.7 became faster and more 
useful than 0.5.

Multiple updates in a day happen when a high priority bug is fixed and the fix 
deployed too quickly turns out also to be flawed. We are not dealing with a 
finished tool here which can be left untouched for years on end. Freenet has a 
very long way to go on security, performance, usability and basic features.
> >> >Every non-pedo
> >> >board on 0.7.5 has been DOS-ed out of commission, or is the target of a
> >> >spam bot.  There is currently no content available outside of cp.  Three
> >> >thousand nodes have been shut down in recent weeks.
> >
> >I do not care about Frost. Frost is broken by design, and is therefore depr=
> >ecated, period. Use FMS or Freetalk. If you don't like the negative trust i=
> >ssues with FMS and Freetalk, write a patch to only do positive trust. IMHO =
> >it would be fairly easy.
> i don't care that you don't care about frost. FMS is also broken by design
> with capchas 99.99% unreadable by humans. 

So suggest improved captchas which are easier for humans to deal with while not 
being easily broken by computers. Good luck with that. Actually in the long 
term we are going to have to find an alternative to CAPTCHAs; even if they 
can't be broken automatically often enough to break Freenet-based chat (I'm 
pretty sure they can be), they can be broken by hand for significantly less 
than a cent each by various Russian companies. The solution is some sort of 
scarcity mechanism, probably darknet-only and based on time limits on each link.

> Freetalk may be useful if someone
> ports it to 0.5 but i can't say for sure because i have never seen it. i
> quit using 0.7 two years ago in one of many storms of total unusability.

You're not aware of the storms about "censorship" on FMS then? Basically, FMS 
eliminates spam by allowing users to assign trust levels to each other, which 
propagate so that FMS identify all non-spammer users. Some users abuse this to 
try to suppress known paedophiles etc. There are alternatives that don't allow 
for "censorship", but they generally mean it will take longer for newbies to 
become visible.

Do you have a solution to Frost being broken by design? It's interesting that 
Frost on 0.5 isn't being spammed, I suspect that's because a) 0.5 is tiny and 
b) the spammer realises that spamming 0.5 won't hurt Freenet proper (0.7).
> >Whoever is doing the DoS probably leaves out the pedo boards on purpose...
> >>=20
> no doubt
> >> Is this even remotely true?  Has 0.7.5 been losing large numbers of nodes?
> >
> >Have a look at the graphs.
> >>=20
> sure, publish them where i can see them

Like I said, not my problem. Find an open gateway, there probably are some.
> >> If so, I'm not surprised at that, only that it's taken this long.
> >
> >No, it has nothing to do with Frost suckage. It was a specific event around=
> > December/January that we are recovering from. (Before that we saw slow gro=
> >wth). It was probably due to bugs and performance problems when introducing=
> > new disruptive code - specifically the new packet format code, which was n=
> >ecessary due to both polling issues and new load management issues, but whi=
> >ch we have still not fully sorted out.
> 0.7 has an amazing number of these 'events'

That's because it's actively developed and pre-1.0! Also because it is 
massively under-resourced.
> >
> >Ac,BYyqgAtc9p0JGbJ~18XU6mtO9ChnBZdf~ttCn48FV7s,AQACAAE/graphs/1035/
> >
> >
> >ns,VGDItiCVzCcWAay51faZzcIfAepzeHpzXYvChlueWYE,AQACAAE/stats/1347/
> sorry those URI's are broken. oh wait, you didn't actually insert them to
> 0.5, you want me to re-install 0.7 to look at your pictures. no thanks,
> been there, done that. i'm still sore from the last abuse i got from 0.7.

*If* this is referring to something other than your other mail about 
censorship, please elaborate.

Attachment: signature.asc
Description: This is a digitally signed message part.

Support mailing list
Unsubscribe at http://emu.freenetproject.org/cgi-bin/mailman/listinfo/support
Or mailto:support-requ...@freenetproject.org?subject=unsubscribe

Reply via email to