#saltr (irc2p) | I2P+ 2.7.0-24+ >> i2pplus.github.io | skank.i2p | git.skank.i2p | purokishi.i2p | vituperative.github.io/i2pchat | notbob.i2p | ramble.i2p | cake.i2p | shinobi.i2p | /msg an op for voice

dr|z3d hi radakayot, welcome to #saltr

orignal guy who is going to implement own i2p router? ))

radakayot hello dr|z3d.

radakayot haha, yes orignal.

dr|z3d *** pokes orignal ***

orignal how is it going?

orignal where you able to establish an NTCP2 session?

radakayot it's going really good thanks to your implementation.

radakayot actually I did. now implementing dns server to resolve i2p domains to pseudo ipv6 addresses.

orignal so no new implemntation of i2p router

radakayot i had to borrow total of 13 files from your implementation of course.

orignal only 13? I doubt that's all

radakayot not yet, after the PoC, I'll slowly reimplement them in c.

orignal you need transports, you need netdb, you need crypto

radakayot oh sorry. you are right.

radakayot 33 total.

orignal 33 sounds more resonable

radakayot sent you the file list.

radakayot I patched some methods to make it more suitable for my needs. it's all automatic.

radakayot also removed meshnet support unfortnately.

radakayot i'm hoping you are not planning to make drastic changes to the libi2pd implementation for the next 3 months :)

radakayot but orignal, without your efforts, I couldn't even dream to make a PoC in this short period of time. so thank you again.

radakayot I may have some questions regarding the implementation differences between i2p java and i2pd.

radakayot is there any document that I can find the implementation differences between two projects? (only i2np and i2cp is enough)

dr|z3d I don't think anyone's written up the differences.

radakayot from what i've seen gost support, floodfill handling and congestion controls are different.

radakayot my goal is to be 100% compatible with the i2p java project and specs in the future, that's why I'm asking.

dr|z3d compatiblity is one thing, feature support is entirely different.

radakayot first goal is 1 to 1 compatibility. including ciphers.

dr|z3d if you're coding in C, i2pd is a fine reference point. there are differences in implementation, but compatiblity isn't an issue. the only things that aren't compatible are the gost crypto and support for yggdrasil, which are both exclusive to i2pd.

radakayot as for the features, i don't have any plans to support sam, tunnels, addressbook, proxy etc.

radakayot dr|z3d: I'll leave the gost support for the sake of fsb. :)

dr|z3d don't get orignal started :)

radakayot I see his point actually, Occam's razor.

radakayot ah also, i really want to ask something to you orignal. do you have any testnet setup you can share that i can quickly get it up?

orignal you can drop gost. it's not used anyway

orignal no we don't have a testnet

dr|z3d are you planning to drop gost, orignal?

orignal but it can be configured easily

orignal no, why should I

dr|z3d <orignal> you can drop gost. it's not used anyway

dr|z3d your words.

orignal well you can still set signaturetype=9 or 10

orignal nobody uses it but still can be used

dr|z3d if no one's using it, why not drop it, reduces maintenance burden, streamlines codebase.

orignal it's just slow vs. EdDSA

orignal myabe people still use it, idk

dr|z3d you could sunset it, give people advanced warning. up to you.

orignal goiing to make it configurable

orignal but never have time for it

zzz dr|z3d, would you please weigh in on git.idk.i2p/i2p-hackers/i2p.i2p/-/issues/492 not sure if we should change things or not

dr|z3d what do you want to know, zzz?

dr|z3d if you're generally happy with your collapsible implementation, then you probably want to address your issue by collapsing the server throttling section (top level),

dr|z3d if the structure is the same as it was when I migrated everything over to tables, then you should have a container td for the throttle section. hide/show that.

dr|z3d so server throttling would persist as a your toggle, and the throttle table (and the container td) would be hidden/shown.

dr|z3d to differentiate between the toggle and the inbound/outbound connection limits, indent the th headings perhaps.

dr|z3d from a UX perspective, forcing the user to uncollapse sections to review configuration options isn't ideal for the most part, but I don't think we're discussing that per se.

zzz pls read the ticket carefully. I'm asking your opinion on the (unexpected?) presence of a 2nd-level collapsible element, is that confusing or not, if so what's the best way to fix it

dr|z3d I just read the ticket, please read my reply carefully :)

dr|z3d tldr: collapse the server throttle section, not the children. indent th titles.

dr|z3d that's the only section in the + tunnel manager that receives a toggle, and the toggle state is persisted.

zzz it's a UI design question. implementation is irrelevant

dr|z3d right, and I'm advising you how to fix it.

dr|z3d do not apply toggles to anything other than top level sections.

zzz I reread your reply, still not sure what you're saying. Is a 2nd level collapsible section (requiring a 2nd click to open it) confusing or not?

dr|z3d it's wrong.

dr|z3d you are providing section level toggles. so nested toggles are bad. the throttler is a single section, treat it as such.

zzz I think I agree

zzz would appreciate it if you briefly commented in the ticket, as eyedeekay isn't here atm

dr|z3d I don't have any skin in the game (your tunnel manager UI choices), but feel free to summarize what I've told you here.

zzz was just looking for a 2nd opinion to push us one way or the other

zzz I'll just paste in the above

dr|z3d roger that

zzz thx

dr|z3d no worries

dr|z3d I think eyedeekay's probably learnt a thing or two since he implemented those toggles, but as a general rule, if you're implementing collapsible sections solely to hide content that's not presented optimally to the user, don't!

dr|z3d collapse sections for lesser-used features that don't need to be instantly available, sure, but if by collapsing sections you're making it harder to review and configure options, you haven't actually fixed anything, you've probably made it worse.

zzz better to have a design philosophy up front, but seems like we're backing into one post hoc

zzz in other news I have a .i2p.alt MR up for review and testing

dr|z3d how does that work? you request i2p.alt and it redirects to .i2p host via the http proxy?

zzz no redirect, it just resolves

zzz same as how www.stats.i2p works

dr|z3d ok, sorry, that's more or less what I meant. so the .alt gets stripped automagically?

zzz which reminds me, not_bob_afk should probably not accept www.* registrations

dr|z3d notbob isn't a registrar.

zzz yeah it's stripped at the naming service

zzz than maybe he shouldn't accept it from reg.i2p? or I guess our complaint is with reg

dr|z3d yup, blame reg.i2p

zzz so the MR is just recognizing that *.i2p.alt is an i2p address, don't send it to the outproxy

dr|z3d sounds good

zzz hmm I don't see www.bocheng.i2p on the reg latest page, but it does work on its jump page

zzz on notbob, bocheng.i2p and www.bocheng.i2p are both listed with the same b32

dr|z3d maybe he could filter those in the UI where the b32 is the same.

zzz dunno what i2pd does with it, maybe worth a discussion

dr|z3d but ideally you'd probably want to have a word in R4SAS's ear and request that he doesn't permit www. registrations.

zzz the wierd thing is that the guy tried to register www.bocheng.i2p with me with a _different_ dest than for bocheng.i2p

zzz I tried to jump to www.stats.i2p on reg and it said not found

dr|z3d yeah, probably because he tried and failed to register both using the same b32?

zzz I don't allow www.* at all

dr|z3d now seems like a good time to have that chat with R4SAS, firm up some global policies.

zzz first let's ask orignal if he strips www from lookups

not_bob_afk I see 27 sites that have been registerd wiht www.* at some point.

not_bob_afk None of them are alive from what I can see.

zzz I only strip it after looking it up first, so I do allow www. in local addressbook, just not on stats.i2p

zzz not_bob_afk, the recent one is www.bocheng.i2p and bocheng.i2p, both on your site with same dest

zzz didn't come from me, I rejected the www one

not_bob_afk I stand corrected. www.mail.i2p is alive

not_bob_afk Ahh, yes. That one has poor health.

not_bob_afk So, to be clear. You never want www.*.i2p to work, ever, right?

not_bob_afk www.infoserver.i2p is also alive.

zzz I think that's the recommendation, but you could leave existing ones in there

not_bob_afk Hmm

not_bob_afk I can easily make my scanner just die if it sees that pattern.

zzz re: www.mail.i2p and mail.i2p and *.mail.i2p, those should probably be removed/rejected as reserved hostnames

zzz they probably aren't controlled by postman and will just lead to confusion

dr|z3d better to just filter the www. site from the listings if it has a duplicate with the same hostname sans www.

dr|z3d *same hostname and dest

zzz I also don't see www.bocheng.i2p on inr.i2p, it seems like it's getting filtered in the reg and/or inr UI

dr|z3d since we're loosely on the subject of addressbooks, what about fixing up the delete hosts function? :)

zzz assuming reg.i2p is where it was registered

zzz whats broken about it

not_bob_afk I would love to have a funtion where I can call and delete a host entry from my addressbook programaticly.

dr|z3d not sure identiguy is accepting registrations currently. so that leaves dnschudo.i2p (?) or isitup.i2p ?

zzz you can't access the addressbook blockfile directly when i2p is running because there's no locking, you could corrupt things

dr|z3d I was thinking more something at the UI end of things.

dr|z3d Delete host from addressbook, it gets moved to a new deleted addressbook with the option to block when requested (proxy error presented to user).

zzz your choices are either to stop i2p first and have drz give you a CLI command, or have drz give you an ad hoc API with a POST form thru the console, or extend i2pcontrol, or or or

not_bob_afk Hmm

dr|z3d currently you delete a host, it gets restored when subscriptions are updated.

dr|z3d so delete is functionally *almost* useless.

zzz shouldn't, if you're using etags or last-mod, and the subsciption server only sends you the new ones

not_bob_afk Unless the site in question is no longer in any hosts files that are in your subs.

zzz if you get the whole thing, not just the new ones, your subscription server is semi-broken

dr|z3d we should probably work on the assumption not all servers are functioning correctly and handle deletion locally as well, perhaps.

not_bob_afk I would like the option to auto delete hosts after a period of time when they are dead to free up the hostname again. Say, a year or the like.

dr|z3d we've had this discussion before.

not_bob_afk I also have some cruft in my addressbook of sites that have never worked.

not_bob_afk I agree!

dr|z3d it's not like we're running out of names.

not_bob_afk We have.

dr|z3d let's focus on one thing at a time. deleted hosts to new addressbook. :)

zzz what's your requirements not_bob_afk ? delete only while i2p is stopped, or while it's running?

not_bob_afk zzz: While running.

zzz ok, so you can't do it with CLI, you need to talk to the running i2p instance

not_bob_afk zzz: Right

not_bob_afk Unless the cli can talk to the running i2p instance.

zzz so your choices are some POST thing via the console, or i2pcontrol extension (json), or maybe a SAM extension

dr|z3d as zzz mentioned, curl/wget with a post request would probably suffice. I can't see any form nonces on the page.

not_bob_afk Ok

zzz there is a nonce in there, at leat in canon

zzz <input type="hidden" name="serial" value="446871204712081361">

not_bob_afk YUes, there is.

dr|z3d ah, missed that, I was looking for a nonce property.

zzz so it's whatever you and drz want to cook up

zzz assuming you can handle a GET for the nonce (or not) and then a POST to some new form

not_bob_afk Doable, but yes.

zzz i2pcontrol sounds harder for both of you

not_bob_afk It's been on my wishlist for a while now, but it's been low priority.

zzz the java code for i2pcontrol is unpleasant

dr|z3d if I were attempting it, I'd probably do something with javascript to fetch the page, extract the nonce, then take the hostname arguments and submit then via post.

dr|z3d haha, unpleasant.

zzz i guess a SAM extension is another possibility, although as a general feature that's pretty insecure

zzz you could probably just POST to the existing page, submitting one or more "checked" params, it probably doesn't even look to see if the values were on the page

mareki2p Eehh, I don't want to be "that guy" again. Do you guys have any automated CI? The I2P.Plus repo fails with the ant i2psnark command, the i2p.i2p repo is fine.

mareki2p BUILD FAILED

mareki2p Blinded message

mareki2p Cause: The name is undefined.

mareki2p Action: Check the spelling.

mareki2p Action: Check that any custom tasks/types have been declared.

mareki2p Action: Check that any <presetdef>/<macrodef> declarations have taken place.

mareki2p On latest Ubuntu

mareki2p Ubuntu 24.04 LTS x86_64

mareki2p dr|z3d: ?

cumlord not_bob_afk forgot about this, managed to get it working :)

cumlord i can put it in a quick cli thing if you want, not sure how you want to use it