@RN
@RN_
@StormyCloud
@T3s|4
@T3s|4_
@eyedeekay
@orignal
@postman
@zzz
%Liorar
+FreefallHeavens
+Xeha
+bak83_
+cancername
+cumlord
+hk
+profetikla
+uop23ip
Arch
DeltaOreo
FreeRider
Irc2PGuest19353
Irc2PGuest46029
Irc2PGuest64530
Meow
Nausicaa
Onn4l7h
Onn4|7h
Over1
acetone_
anon4
anu
boonst
dr|z3d
enoxa
mareki2pb
mittwerk
not_bob_afk
plap
poriori_
shiver_
simprelay
solidx66
u5657_1
weko_
zzz
dr|z3d, while I appreciate the testing and enthusiasm, further testing over here has revealed some robustness and performance issues with persistence
zzz
I believe it's actually making things slower
dr|z3d
hmm.. what's leading you to that conclusion?
zzz
so I recommend you disable it for now
zzz
it's going to take some time to work through the issues
zzz
I2PTunnelHTTPClient.java: boolean keepalive = !(s instanceof InternalSocket);
zzz
change to either:
zzz
boolean keepalive = false;
zzz
or if you want an option:
zzz
boolean keepalive = ctx.getBooleanProperty("yourchoice") && !(s instanceof InternalSocket);
dr|z3d
ok, thanks. not seeing much in the way of slowdown here, are you testing specific sites or?
zzz
it's (sometime) stuck waiting for the i2p socket to close before going around again, which is adding several seconds of lag
dr|z3d
anywhere we can log that stuckness when it happens?
zzz
I promise it's happening :)
dr|z3d
I believe you :)
zzz
basically it's the time between closing the last req/response and reading the next one
dr|z3d
we're not using a thread pool right now, are we?
zzz
there's also quite a few corner cases and stuff I'm working through, some fixed locally, some not
zzz
there are thread pools involved in some places but not others
zzz
if you want the logs it looks like the browser waits 3 sec before sending another request, which wouldn't seem unusual
zzz
but it really just took us 3 sec to get around to reading the request
zzz
you can't really see it without hacking in a bunch more logging which is what I have locally
dr|z3d
ok, well, if you want to push that logging to your MR, maybe I can help debug, if not, I'll probably just disable keepalive as you suggest.
zzz
if I were in that branch all day I could iterate fixes like that but unfortunately I really need to stay in canon to keep up with and help eyedeekay
zzz
and its not practical for me to flip branches back and forth several times a day
dr|z3d
ok, np.
zzz
the overall design is still sound, it's just not ready and may not be for a while
dr|z3d
sure, no rush, just keep me in the loop when you think you've fixed the current issues :)