zzz
remote: rpc error: code = DeadlineExceeded desc = waiting for packfile negotiation: context canceled
eyedeekay
Which remote?
zzz
i2p.plugins.firefox, died after 25 MB. What in the world did you check in there?
zzz
and, unrelated, why is every javadoc file a separate checkin?
zzz
I know you're working on cleaning things up but still seems very messy
eyedeekay
The script does (generate .md from javadoc)->(generate .html from .md)->(checkin change) for every Java file, I can make it so that it holds off on the checkin until it runs out of java files though
zzz
yeah, run your operation like people are watching
eyedeekay
Re: the big files, that would be the extension zips, those will be gone when I switch to making every build also generate the static resources
zzz
was I getting close at 25 MB? or should I give up and pull from github?
eyedeekay
Pull from github for now and I'll yank out the zips tonight
zzz
dunno if you can find a server-side setting for 'deadline exceeded', put it on the todo list
eyedeekay
Can do
zzz
I wanted to spend a quick 5 minutes skimming your changes, but the pull died and I gave up scrolling past pages and pages and pages of javadoc revs. Make it easier for people to help you ))
dr|z3d
I stumbled across a strange side-effect of corrupting udp packets earlier.
dr|z3d
It turns out that you can publish a routerinfo to the netdb that is reachable for other routers without any traces of an ip address.
dr|z3d
zzz: it _might_ be worth the effort to validate routerinfos and temp ban any without an attached address if they're not being introduced?
dr|z3d
On a related note, I'm temp banning routers that have no routerinfo, though I'm not sure they exist in the wild.
zzz
aren't those called hidden routers or routers with introducers?
dr|z3d
no
dr|z3d
the router appears to other routers as reachable.
dr|z3d
that's the odd thing, reachable, and yet without any ips, no introducers.
dr|z3d
there's another hoop to jump through to get there.. no udp peer tests.
dr|z3d
I should mention that ips are absent from the local router info, too.
zzz
all routers have router infos, but that doesn't mean you have it
dr|z3d
sure, but if I look at a router in the netdb, and absent ips, all the info is present, and it's marked as reachable, then I have said routerinfo :)
zzz
"marked as reachable" how?
dr|z3d
it appears in the netdb with the 'R' flag displayed.
zzz
that's the router's opinion of itself, not your opinion of it.
zzz
if it's publishing R with no inbound addresses, that's a bug, but not a big issue
dr|z3d
it's likely an artifact of a bug I introduced, corrupting udp packets. but it was curious nonetheless.
dr|z3d
the R status is a side issue, in any event. the main thing was the "hey look, mom, no ip addresses, no introducers!"
dr|z3d
I was asking myself if that could be leveraged for hidden routers somehow, or if it could be abused.
zzz
those are hidden routers, or, perhaps, routers that don't know their IP yet
dr|z3d
not hidden, like I said, was appearing in the netdb of other routers. and the ip addresses were absent locally as well as remotely. persistent throughout the session of the router in question.
zzz
I've lost the thread of whether this is your bug or something you've seen in the wild or just philosophy, but it can happen
dr|z3d
router in question limped along, but was severly impeded in general performance, part tunnels especially. issue was specifically my bug corrupting udp packets.
dr|z3d
in the absence of any actual benefit of publishing a "reachable" routerinfo to the netdb without any ips or introducers, I'm suggesting that maybe temp banning said routers if they appear.
dr|z3d
I guess that's the tldr.
zzz
not worth the trouble imho
dr|z3d
ok, fair enough.
dr|z3d
interesting peer summary page, zzz.
dr|z3d
the version column could go. just suffix the transport with the version. NTCP v2, SSU v1, SSU2 v2.
dr|z3d
The Total column probably wants to go on the far right.
dr|z3d
in and out columns could also be merged.
dr|z3d
totals would then read 3 / 6 or whatever per ipv column.