@eyedeekay
&kytv
&zzz
+R4SAS
+RN
+RN_
+T3s|4
+dr|z3d
+hk
+orignal
+postman
+wodencafe
Arch
DeltaOreo
FreeRider
FreefallHeavens
Irc2PGuest15271
Irc2PGuest28511
Irc2PGuest64530
Irc2PGuest77854
Nausicaa
Onn4l7h
Onn4|7h
Over
Sleepy
Soni
T3s|4_
Teeed
aargh3
acetone_
anon4
b3t4f4c3
bak83
boonst
cancername
cumlord
dr4wd3
eyedeekay_bnc
hagen_
khb_
not_bob
plap
poriori
profetikla
r3med1tz-
rapidash
shiver_1
solidx66
u5657
uop23ip
w8rabbit
weko_
x74a6
dr|z3d
zzz: another possible usecase for restricted routes, one which I'm contemplating right now on account of a server that's super-locked down.. inbound and outbound ports can only be opened on an individual basis.. so nominating specific servers as the first hop presumably would mitigate that?
dr|z3d
not entirely sure that would fix the issue, but not sure what else would.
dr|z3d
also, noticed you're handling the i2pd key import now.. throw new InvalidKeySpecException("todo"); <-- maybe something other than todo as the exception error? :) "Unsupported key type" ?
zzz
the todo won't be seen by the user, it's caught in the loop now, so it will fall through to the unsupported key type exceptino
dr|z3d
ok, no problem then. thanks.
zzz
I saw you bypassed merging the translation updates rev, on purpose, or you need a better process to not miss things, or you're not supporting translations any more, or?
dr|z3d
mostly because upstream translations are becoming less and less relevant for I2P+
zzz
ok, so it's going to gradually become english only
dr|z3d
no, the existing translations are good, manually submitted .po updates are accepted, and I'm diving into poedit myself to do some stopgap translations.
zzz
ok
dr|z3d
there are a lot of new strings appearing in I2P which mostly don't appear in I2P+, so pulling translations every cycle doesn't make much sense.
zzz
if you werent doing your own translations, you could just take ours and get as much as you can out of it, but since you are, you'd have to use gettext tools to merge them somehow
dr|z3d
yeah, merge hell in essence. translations aren't in a bad shape, in any event, in some cases better than I2P, coverage wise. so it's all good :)
zzz
idk put up some machine translation service eepsite, with some scripting you could probably translate the whole corpus for all languages, who knows what the quality would be
dr|z3d
the quality would be pretty abysmal. :)
dr|z3d
I know, I've used google's translate on website text, and the feedback is always along the lines of "awful" :)
zzz
sure for the longer sentences, but you'd get a lot of impact just from restricting to e.g. 1 or 2 words
dr|z3d
that said, for the target audience, badly translated text might inspire someone to do the work to make it good.
dr|z3d
and sure, true, keyword translation might have a better signal to noise ratio.
zzz
that was always my theory, almost any translation is better than nothing
zzz
re: server lockdown, are you talking about destination ports for outbound sockets/packets that's part of the equation?
dr|z3d
yeah, dest ports, indeed.
dr|z3d
server's locked down ridiculously tight. would like to get something working, it's on a 10Gb pipe.
zzz
but there's some facility to whitelist?
dr|z3d
ports can be whitelisted individually, sure.
zzz
so whitelist every destination port 9151-30777
dr|z3d
lol
dr|z3d
when I say individually.. I mean *individually*
dr|z3d
you can't specify a range.
zzz
for i in 9151..30777; do; whitelist $i; done
dr|z3d
no, it's not a standard linux firewall.
dr|z3d
there is no firewall active on the system itself, it's one level above.
dr|z3d
and the one-level-above firewall is a GUI app, no option to whitelist more than one port at a time.
zzz
then yes, that's a use case for restricted routes
dr|z3d
ok, thought it might be, wasn't sure. and we just specify comma separated ips for the nominated routers?
zzz
restricted routes is not implemented
dr|z3d
Oh, I thought you said you'd tested shouldSelectExplicit() previously. my bad. or maybe that's something slightly different.
zzz
slightly different and janky
dr|z3d
well, 10Gb/s router would love to have access to restricted routes :)
zzz
more of a test mode
dr|z3d
ok
zzz
well, jrandom said it would happen in 3.0, we're not there yet
dr|z3d
lol
dr|z3d
3.0, sometime in the text millenium.
dr|z3d
*next
zzz
the ancient thread on my forum lists some of the things R.R. _could_ mean, and _might_ require
zzz
its really just a handwaving catchall term at this point
dr|z3d
yeah, looked at that. well bypassing restrictive outbound firewall policies, definitely top of my list right this moment.
zzz
but without full R.R., wtf good is 10Gbps if a) it's all going to a handful of presumably much slower routers, and b) nobody can build a tunnel through you?
dr|z3d
well, this is the question I'm asking. if the router could nominate a proxy, as i2pd permits, that's half the issue fixed. sure, the bottleneck then is the upstream proxy, but 10Gb/s is never going to see full usage anyways, same for 1Gb/s, so I dunno. just trying to work out some strategy to get things moving in the right direction.
zzz
hidden mode + 1 hop tunnels + explicit peers will get you started
zzz
if you want longer tunnels you'll have to change how explicit peers works to put your favorites at the closest hop
zzz
and not use them for the farther hops
zzz
but >1 hop tunnels is going to be more fragile/janky
zzz
because there's a bootstrap problem
dr|z3d
hmm, as you say, janktastic. :|
dr|z3d
I guess what would be useful is an outbound proxy option, since inbound traffic is fine, or the option to nominate routers to act as proxies for outbound traffic.
zzz
a "proxy" network role (IB or OB) would be some brand new network architecture thing and a huge amount of work
dr|z3d
yeah, figured as much. :|
zzz
to the extent things can be accomplished with existing protocols its a lot easier
dr|z3d
well hidden mode doesn't achieve much more than a foot in the door of the network. not really a viable solution.
zzz
thats why hidden mode + closest-hop restrictions are more viable
zzz
sure but why would anybody want to connect to or route a tunnel thru such a router?
dr|z3d
hidden mode, no good, not if you want to share bandwidth. as it is, router is hosting 25 part tunnels and seems to think it's OK, status-wise.
zzz
if you're using your friends as closest-hop, then a tunnel thru them and you as colluding routers is undesirable
dr|z3d
if I can delegate alice to handle all my outbound traffic, what would that be, reverse hole punching (?), then I can function more or less normally as a router.
dr|z3d
well, not "alice", but routers I trust/own.
dr|z3d
but, sure, it's problematic on the collusion front I guess.
zzz
you're useless as OBEP or IBGW, and as middle hop only with two of your friends on either side?
dr|z3d
i2pd allows you to nominate a proxy to route all your traffic through. but you're suggesting that's a huge amount of work. is it useful in this context? dunno..
dr|z3d
yeah, router x won't want to route through router y just to talk to me. so hmm.
zzz
we don't bother trying to route tunnels through hidden mode routers and usually avoid firewalled routers too. Are they some great untapped resource that would benefit the network?
dr|z3d
except via introducers, which is a different proposition.
zzz
why bend over backwards trying to use some other type of handicapped router
dr|z3d
the answer to that question is "probably not", rhetorical though it may have been :)
dr|z3d
otoh, there are probably a not-insignificant amount of routers out there that could be contributing more were it not for restrictive outbound firewall rules.. campus dorm routers etc.
zzz
i2pd-type proxying is feasible, yes a lot of work, but for now like with ygg we leave that problem space to i2pd
dr|z3d
maybe restricted routes in this context is the wrong focus. a proxy approach is perhaps more useful, yes.
dr|z3d
not pushing for a proxy approach, anyways, just raising the issue, hopefully not too forcefully.
zzz
the network architecture is that everybody wants to talk to everybody, so as soon as that's not true (your router, ygg-only, ipv6-only, ...) it gets ugly in a hurry
dr|z3d
very true.