@eyedeekay
+R4SAS
+RN
+RN_
+T3s|4
+Xeha
+acetone
+orignal
+weko
Irc2PGuest5995
Irc2PGuest89954
Leopold_
Onn4l7h
Onn4|7h
T3s|4_
aargh2
anon2
eyedeekay_bnc
hk
not_bob_afk
profetikla
shiver_
u5657
x74a6
zzz
ping dr|z3d
dr|z3d
*pong*
zzz
ok
zzz
first
zzz
dr|z3d, since I now have to ask this every time, is irc.postman.i2p running on plus?
dr|z3d
you'd need to talk to postman. not intimately acquainted with the topology of his infrastructure.
zzz
ok will do
zzz
second
zzz
the infinite loop through the netdb lookup code is extremely concerning
zzz
so I need to put this on the table
zzz
the possibility that plus is contributing to the network issues
zzz
how likely? dunno
dr|z3d
not very likely is my guess, if at all. I'd have seen the infinite loop bug on other routers if it wasn't pretty rare.
zzz
but your netdb structural mods are ill-conceived, borked, and glitch-prone at best
zzz
I recommend that you back out all of it
zzz
if you don't agree, then perhaps what I need to do to protect the network is drop everything and walk you through all the issues
zzz
which really isn't what I want to or should be doing
dr|z3d
in any event, users updating from either in-console the release or dev urls have the fix.
dr|z3d
*have the fix available
dr|z3d
that's a bit too broad brush, I have no idea what you think is ill-conceived, borked, glitchy.
zzz
you're making significant changes in code you really shouldn't be mucking in
dr|z3d
I'm still none the wiser.
zzz
right because it's not my yob to worry about downstreams, but if they threaten the network I guess I have to
dr|z3d
I think you're overstating the threat tbh.
dr|z3d
based on a bug that's now fixed that I've never seen on a fairly sizeable chunk of routers, only via a report.
zzz
maybe, but we have seen what a small number of routers can do
dr|z3d
ok, let's be clear. i2p+ routers aren't responsible for a) oversized b/w requests or b) excessive tunnel requests.
dr|z3d
those are the 2 issues that are currently afflicting the network.
dr|z3d
I'm happy to take guidance where you think I've taken a wrong turn, but telling me my code is borked and whatnot isn't very helpful.
zzz
probably not, but we don't know what other contributing causes may be out there
zzz
right, I know I'm not being helpful, I'm just frustrated
dr|z3d
well, yeah, exactly. we don't know.
dr|z3d
so it's a stretch to start hypothesizing that i2p+ is the issue.
dr|z3d
why don't we work from the basic hypothesis that the network's under attack?
dr|z3d
that's a good starting point, and one that's a lot more credible.
zzz
just putting it on the table, not assessing a likelihood. But your mods, like the 20-introducer mod guys, are not benign, and we need to classify your product as a separate implementation
zzz
and treat your product as an individual suspect, as the other implementations are
dr|z3d
what's not benign?
dr|z3d
if you're really suggesting my mods are hostile to the network, you should run i2p+ somewhere so you can validate your thesis.
zzz
making changes that can infinite loop? that affect fundamental code?
dr|z3d
otherwise, it's just random conjecture, and it doesn't give me anything actionable. I've already fixed that loop, as I mentioned on zzz.i2p, and apparently it works according to the reporter.
zzz
I'm saying your mods have that potential. Not a thesis.
dr|z3d
yesterday it was all about "we all make mistakes, shit happens". today you're accusing me of non-benign mods. I don't think that's even remotely fair.
zzz
non-benign as in they're not inherently safe
dr|z3d
I've fixed the infinite loop bug, that was unfortunate. the commit is on gitlab if you want to review it for sanity and the like. but I take issue with your claims that my code is borked. If you find stuff that's genuinely borked, I'm always listening.
zzz
I'm not accusing you of anything. I'm assessing the quality of your code and its potential to spiral out of control
dr|z3d
I can't work with vaguaries or generalizations, notwithstanding.
zzz
sure, I get it
zzz
I have to decide if this is how I want to spend my time
dr|z3d
If I had your concerns, I'd run I2P+ in a VM or whatever and just keep an eye on it for a while, run it with extending logging, and see if it has the potential the blow up the network :)
dr|z3d
I think your concerns are valid, but overstated, but I'm not looking to dismiss them out of hand.
dr|z3d
Even if I didn't have your concerns, I'd probably be curious to learn how I2P+ functions relative to I2P, but that's just me :)
zzz
thank you. my only message for now is code carefully. If I can gather the energy and time to do a code review, I'll try to schedule some time with you
zzz
but I won't be running your router with a ten foot pole right now :)
dr|z3d
entirely up to you, no obligation :)
RN
I'm running one if I can test anything or help in any way
dr|z3d
re coding carefully, that stackoverflow bug was from fairly old code in the scheme of things. I'm more circumspect now about what I'm modding, at least I hope I am.
dr|z3d
if you see a stackoverflow bug, RN, holler. unlikely, should be fixed, but any ERROR or CRIT bugs are noteworthy.
zzz
the reason it's working is you have at least two bugs on top of each other that are cancelling each other out, so don't think everything is ducky now
dr|z3d
ok, I'll review the code some more. thanks.
zzz
thats why I don't want to get into it wihtout a lot of time because it could get a lot worse if done halfway
zzz
you may wish to review and assess which of your significant changes over the years are still necessary today
zzz
then make a list of those and ask for eyeballs
zzz
and revert the rest to minimize risk, test time, and support time
zzz
you're in a bind because nobody is looking at your code. ever.
dr|z3d
I get some code review, though maybe not quite at the level I'd like, and I also get out-of-band bug reports. not as bad as you're suggesting.
zzz
ok
dr|z3d
I appreciate the offer of "eyeballs" though. Thanks.
dr|z3d
just because we're on separate code bases doesn't mean we can't help each other :)
zzz
well, hopefully not just mine
zzz
but you'd have to tell us where to look
zzz
my main concern is netdb but even within that is 51 separate files
dr|z3d
that's why there aren't too many sets of eyeballs competent to review the code :)
zzz
and 12,852 lines
zzz
on my side
dr|z3d
if there are bugs obvious to you in IterativeSearchJob.java, they're not so obvious to me. the intention there is to keep the router from refreshing routers that we're not going to be using for local tunnel builds.
dr|z3d
canon just reduces the expiry for routers to keep the netdb in check, cannon attempts to grade routers based on reported performance and expire the crap earlier.
zzz
and because you changed all the log lines there's no way for anybody to easily audit your changes
zzz
thats why you'd have to point people in the right direction
dr|z3d
necessary. canon formats the logs that are probably useful to == 1 person :)
dr|z3d
when I read logs, I don't want to have to decrypt obscure acronyms or scratch my head trying to parse multiple lines of unformatted text.
zzz
i mean shouldLog() -> shouldxxx
dr|z3d
don't know what you mean. talking about the level of logging events?
zzz
tweaking everything just adds unnecessary risk to your users, and yes, to the network
dr|z3d
oh, that. the first is just longhand for the latter, no?
dr|z3d
a for reverting, it's all about what I'm trying to accomplish. I prefer to refine not revert unless something is evidently broken.
zzz
there could be, for example, anonymity/security risks to your users that don't affect the network
RN
dr|z3d, grepped all logs, no instance of "Stack"
zzz
so for each change me or you or anybody makes, it introduces some risk epsilon and takes time, and it's our job to assess that it's worth it
zzz
and unique to you, it increases your maintenance burden by a different epsilon
dr|z3d
current tunnel build success on a fairly loaded system: 73%. do I win a prize?
dr|z3d
can't be all bad, zzz :)
zzz
if that's expl + client, I can't line that up w/ anything, all of stats.i2p is expl only
zzz
dunno what i2pd does
dr|z3d
yeah, it's total build success. I figured that's a more useful metric that just expl or client.
dr|z3d
added a graph as well, so easy to track performance over time.
dr|z3d
could be good to do a client + expl percentage graph on stats.i2p
zzz
not a good time to be offering me suggestions/couldas/shouldas, I need a couple weeks
zzz
in general. not about you :)
dr|z3d
ok, what's your focus right now?
dr|z3d
obviously something's got under your skin :|
dr|z3d
if it's not i2p related, forget I asked.
zzz
a lot going on and too many interrupts
zzz
so if I sound cranky, I am, again, not about you
zzz
I just need a minute
dr|z3d
take all the time you need
zzz
thanks for your understanding. I rarely cry uncle but now's the time