#dev (ilita) | Обсуждение разработки I2Pd | i2pd development discussion | http://i2pd.website/

R4SAS вижу хрень в андроид коде

R4SAS github.com/PurpleI2P/i2pd-android/blob/master/app/jni/DaemonAndroid.cpp#L39

R4SAS мы тут прописываем путь до datadir

R4SAS потом ниже вызываем Daemon.init

R4SAS и потом снова ставим пусть на автоопределенный

R4SAS github.com/PurpleI2P/i2pd/blob/openssl/daemon/Daemon.cpp#L77

R4SAS крч тупость

R4SAS думаю чего делать

R4SAS хотя идея есть

R4SAS orignal: в Daemon_Singleton создавать функцию через которую буду указывать для синглтона параметр папки - это хенрня?

R4SAS почему там - потому что в init уже идет проверка папки и создание подпапок. вызывать установку папки раньше нельзя ибо оно не будет в init учитываться, и после init не создаст нуждные папки

R4SAS orignal: paste.i2pd.xyz/?1877392078bcfcdb#A48aMTyfRuL6QPLb9TBFFb37tq6ztEw9ABCjrDGDm2Ey

R4SAS вот так

orignal не понял вопрос

orignal да так нормально

orignal я бы тоже так сделал

orignal zzz you send fgagmented I2NP with RouterInfo and always fails on decompression

orignal Block type 5 of size 637

orignal and Block type 4 of size 1192

zzz hmm

orignal do you know actual i2np message size?

orignal I'm trying to found what's wrong

zzz Received partial ack of 289617048 by [Hash: cdoFbNTTgmjfUlL6M8BYQTGdq2QxGsAa7xkX5xnfW4o=] after 138 and 1 sends complete? false newly-acked: 720 fragment: 1 for: OB Message 289617048 seq 0 type 1 size 1824 fragments: 2 volleys: 1 lifetime: 138 unacked fragments: 0 sizes: 1192 632 send counts: 1 1

zzz sorry dup paste

zzz no I don't have the i2np size logged

orignal what is 1824?

zzz well, it's 1192 + 632

zzz =1824

orignal then why I receive 637?

zzz that's the 5-byte overhead for frag + msg id

orignal 637 is size field of that block

orignal right

orignal and 1192 is with header

orignal yes my pyaload size is 1815

orignal seems right

zzz right now I have extra stuff in my RI to make it big, for testing fragmented session confirmed

orignal might it be the reason?

orignal eveything looks right but gzip fails

zzz failing the checksum I guess?

orignal not sure

orignal it's inside libz

zzz is your fragment handling and reassembly working for other i2np messages?

orignal the problem is I just started it

orignal didn't have other examples

zzz that would be my guess

zzz I'll look in my logs to see if I see anything wrong on my side

orignal have to tried to handle that RI?

zzz the session confirmed fragmentation is working

orignal great

orignal will implement it

zzz but of course that's different than data-phase i2np fragmentation

orignal ofc

zzz the session confirmed fragmentation has been out in the network since tuesday. you should be seeing session confirmed AEAD failures for inbound connections until you support it

orignal I havn't checked that router yet

zzz can you try fetching my RI from a ff? that will tell you if there's something wrong

orignal should it be the same?

orignal I have database store msg

zzz yes

orignal I mean

zzz i fetched it ok from another router

orignal stange

orignal maybe just fragmnetation bug

zzz that's my guess

zzz gzip checksum catches it, other messages you just pass through

orignal libz direnetly does

orignal R4SAS void Daemon_Singleton::setDataDir(std::string path)

orignal const std::string& path

orignal if (path != "")

orignal if (!path.empty ())

zzz try to hexdump the gzip data on failure, we can look at it

orignal will do

orignal that router seems crashed

orignal will take care

R4SAS orignal: ну... можно и так

orignal да по сути без разницы

orignal просто гвоорю как надо по фэншую ))

R4SAS я понимаю

R4SAS так же подумал перед тем как заливать

orignal да пох

R4SAS теперь у тех кто не мог запустить из-за кривого определения папки должно заработать

orignal на телефоне?

R4SAS да

R4SAS это всякие сяомишники, хоноры

orignal ээээ

orignal надо будет таки телефон прикупить

R4SAS для чего

R4SAS тестов?

orignal ну да

orignal да и вообще я как из другого мира ))

R4SAS про то что на древности сидишь?)

orignal что у меня телеона как из другого мира

R4SAS хорошо что не моторола

R4SAS :D

Конёк_Горбунок_ЇмператорЪЬ_Заїа Книга и фильм "Человек-невидимка"

Конёк_Горбунок_ЇмператорЪЬ_Заїа Фантастика стала реальностью, но иной

orignal это что за высер?

orignal zzz, I have save that gzip portion

zzz see any zeros in the middle or at the end that would indicate a reassembly problem?

orignal and was about to unzip using gzipo

orignal sec

orignal outer.version=^F0.9.53;^Au=<B3>TEST (*M@ECINIHFI

orignal I see shit like this there

zzz that looks right, more or less

orignal eveything else looks right

zzz ^F, for example, is string length 6

orignal yes I know

orignal I mean TEST, etc,

orignal and I know why is it

zzz that's correct, I put it in there to make the RI bigger

orignal my max buffer size is 2K

zzz u = TEST (*M@ECINIHFI(@#FIJG$FI(NG$I(n43i9pfp9ig4n8iINWF(P(*C((*RJGIVNWEUIF)Q@*(!(*)(!&*()#R!(*JF(IHFIOHWFIOHCINOJFIOHFIOHCINWCJVSXMVMXVBINWIONVCIONCEIOOWICOWICNOINVOICIOOIEWNOVINOVIN

orignal for RI

zzz yeah my uncompressed router info is currently 2390 bytes

orignal yes I see

orignal that's why

zzz ok, great, another bug found

orignal it's not a bug really

orignal you never see a router more than 2K so far

zzz maybe, maybe not - I see one that's 2002 bytes right now

zzz did it overflow the buffer and corrupt something else?

orignal тщ

orignal no

zzz good

orignal just decompression failure

zzz I put in a bunch of random chars so it would be two session confirmed packets even when gzipped

zzz RIs will get even bigger when we implement SSU2 introducers

orignal no problem will increase buffer size

Конёк_Горбунок_ЇмператорЪЬ_Заїа инженеры что-то не догоняют, что они там инжинирят

orignal весеннее обострение?

Конёк_Горбунок_ЇмператорЪЬ_Заїа нет, преступная нейтральность

orignal zzz, should I set total fragments to 1 even if my RI is not fragmented?

orignal right now I set 0 to this field

zzz in the session confirmed header?

orignal yes

orignal frag field

zzz right now I'm sending 1, and accepting both 0 and 1 for backward compatibility. The spec says 1

orignal so shoudl I always set to 1?

zzz yes, let's do that

orignal basically to (1, 0) if no fragmnetation

orignal will do

zzz super. but accept (0, 0) for now

orignal fine

zzz then we'll know who needs to upgrade ))

orignal good point ))