logs archiveIRC Archive / Oftc / #tor / 2010 / March / 4 / 1
nsa
or: arma committed revision 21791 (/check/trunk/cgi-bin): make it clear which TorCheck script to look at
or: Peter Palfrader <peter@palfrader.org>: 2010-03-03 22:52:43 [debian-tor/master]: Two debian/rules fixes
or: arma committed revision 21792 (/check/trunk/cgi-bin): make the set of languages more maintainable. no idea if this
or: will be valid python code; somebody should fix it.
or: arma committed revision 21793 (/check/trunk/cgi-bin): switch to html characters; get rid of the flags
or: arma committed revision 21794 (/check/trunk/cgi-bin): these are all the po files that appear sufficiently translated.
NemesisD
anyone know if theres a documented list of error codes and their meanings? i'm working on some bindings for ruby and I want to make a heirarchy of exceptions based on response codes
arma
error codes from what?
NemesisD
arma: oh sorry. from controller signals, etc
it will be like 200 SUCCESS or 500 AUTHENTICATION REQUIRED. I'm assuming that format is consistent and so I want to parse the code and message out
formalist
control-spec.txt?
arma
right, you want to read control-spec.txt
you may also like to look at the python controller, torctl
NemesisD
lol torctl's pydocs are locked
arma
https://svn.torproject.org/svn/torctl/trunk/python/
nsa
or: erinn committed revision 21795 (/torbrowser/trunk): Makefile.linux changes so things build properly, update config files and paths in scripts for TBB Linux.
or: erinn committed revision 21796 (/torbrowser/trunk/build-scripts): fix incorrect path
         

katmagic
NemesisD: oh, wow. I was starting on exactly the same thing a couple days ago.
(I didn't get past reading the spec, but still...)
nsa
or: phobos committed revision 21797 (/website/trunk/en): we have a new address.
Sebastian
outofwords, if you're reading this, all your patches expired. Sorry. I forgot to make a branch out of them right away, and didn't think you'd have them set to expire after 12 hours :(
StrangeCharm
is there a nice copy of the 'onion' logo somewhere about?
ideally an svg, but a hires raster image would rock too. i seem to recall seeing one on the site somewhere, but i can;t recall where right now
atagar
StrangeCharm: I'm sure someone will provide one soon, but tweaked the logo once to make the background transparent in case it helps: http://www.atagar.com/misc/gsocBlog09/resources/torLabel.png
StrangeCharm
atagar, sweet
Sebastian
atagar: there, or-dev day is today
atagar
Gasp! A response?!? One sec...
Sebastian: Nice to see we agree on those. For my purposes lacking client/exit connections will be fine - though I do like showing that 'something' exists so this means it won't be a complete replacement of netstat for me. Would it be satesfactory to have those connections scrubbed (ie, show that *something* exists on that circuit, but not details)?
SwordAngel
Hi guys.
atagar
Sebastian: Gonna be afk for a bit (but please let me know) - if scrubbing is ok then I'll update the proposal this weekend, if not then I'd be interested in discussing it. Cheers!
SwordAngel
I'm using the Vidalia package for Mac OS X and Firefox with the Tor button.
Sebastian
atagar: I think if Tor provides exit/client data - even with an option that is turned off by default and you need a to say please three times to enable it - that's bad.
hi SwordAngel
SwordAngel
With this setup, where does Firefox send its DNS requests? To the usual non-Tor DNS or through SOCKS 5 to the Tor network?
katmagic
Someone could always patch it...
atagar
Sebastian: that's not what I want - I'd like for tor to say "yes, an exit connection exists on this circuit - and that's all we're gonna tell you"
(hence the 'scrubbed' part)
katmagic
Well...
Sebastian
atagar: Hm. Please ask for feedback on that on the mailing list. I think it should be ok, but I'm not at all sure
SwordAngel: via socks 5 to the Tor network
atagar
Sebastian: will do
katmagic
atagar: just a sec. That's pretty easy.
atagar
hu?
         

mib_1jha4e
Port 9051 is blocked/otherwise unusable, is there any other port Tor can use for its ControlPort ?
katmagic
It can use any port it has rights to bind to.
It can also use UNIX sockets.
ControlPort 1337
or ControlSocket /dev/torctrl
And of course, use a password/cookies/socket.
mib_1jha4e
Would 1337 actually work? :P
katmagic
Yes, if you're not already bound to it.
*something's*
mib_1jha4e
Will i have to change what port Firefox connects to Tor from?
arma
firefox doesn't use the control port. only vidalia does, when talking to tor.
katmagic
No, you'll just have to configure your control application to use the new port.
arma
you may be confusing the control port with the socks port.
nsa
or: Sebastian Hahn <sebastian@torproject.org>: 2010-03-04 04:56:23 [tor/master]: Add proposal 170 "Configuration options regarding circuit building"
or: Mike Perry <mikeperry-git@fscked.org>: 2010-03-04 05:02:01 [tor/master]: Woops, forgot the second warn.
or: Nick Mathewson <nickm@torproject.org>: 2010-03-04 05:13:23 [tor/master]: Merge remote branch 'mikeperry/bwweight-smartlistfix'
or: Mike Perry <mikeperry-git@fscked.org>: 2010-03-04 01:35:06 [tor/master]: Demote warn to info, since it can happen.
or: Mike Perry <mikeperry-git@fscked.org>: 2010-02-27 00:17:09 [tor/master]: Check for empty smartlists and no bandwidth during node selection.
atagar
Hmmm, that's odd - katmagic sent me a bit of ruby to list current client/exit connections. Haven't a clue why...
nsa
or: [wiki] RogerDingledine updated TheOnionRouter/TorFAQ - http://wiki.torproject.org/noreply/TheOnionRouter/TorFAQ
total
Hello, user here - I'm in need of some help w/ a "All routers are down or won't exit" problem I've been getting lately
arma
what OS? what tor version? which install did you do?
total
Vista64, Tor 0.2.1.20 and Tor 0.2.1.24, currently running 0.2.1.24
arma
did you change the config at all?
total
no
wait, I thought it was my ISP, so I added in bridges
so yes, I did change the config to use bridges
arma
ah. and you probably have configured a bunch of bridges which aren't actually up anymore?
what happens when you unclick that?
total
hmm..I checked teh bridges website, used those, and also signed up for the bridges emails
lemme check - hold on
arma
is your ISP in the US, or china?
total
US
I get the same spam of "All routers are down or won't exit". If this helps, this happens when I run Vuze 4.3.1.4
arma
ah. well, first of all, you shouldn't be running bittorrent with tor. it ruins the tor network.
there are people in iran trying to get their twitter out, and they can't because of some jerk in idaho who wants another madonna mp3
total
I've only set it up for client - server talk - no peer to peer
erm, tracker communication only
arma
any other interesting warns or notices?
does it work for web browsing via firefox and torbutton?
total
let me check
Tor for FF is working - dumped me in Google Deutsch doing a search for tor
arma
ah. perhaps your tracker is on port 25 or some other port that isn't commonly allowed by tor's exit relays.
e.g. 6881
total
let me check
I've just removed all my torrents in vuze, and I still get the All routers are down or won't exit spam
arma
well, you clearly have something trying to route its traffic through tor to a destination or port that tor doesn't want to reach
i imagine that something is your vuze
total
it probably is something in vuze then
thank you for helping me trouble shoot the problem
now to go bug the Vuze people
outofwords
I am again about empty smartlist and NULL pointer. The last is bug, so "Demote warn to info" for all is wrong. If you think that log can happen to debbug so at least need warn with LD_BUG. Better do tor_assert().
s/happen/helps/
I am about 97eec84f4bcc4908e47ebcb51f1eb1d3a782b42b
arma
the bug does happen. moria1 encountered it.
outofwords
I guess not with NULL pointer
empty list or no bws.
the crash could happened with NULL pointer to smartlist and other (empty or summary 0 bandwith) is different.
the fist could crash in smartlist_len() the last two in crypto_rand_uint64()
the first is the bug of caller if it can happens.
nsa
or: arma committed revision 21798 (/website/trunk/docs/en): putting br inside tt gets complaints sometimes
outofwords
the seconds two is poor checks of func.
About 1270 bug and expired links, I guess the spec violating is not urgent issues, so links will expires after 6 months anyway.
nsa
or: ioerror committed revision 21799 (/website/trunk): Add code to generate QR code for Android and a sample image
or: ioerror committed revision 21800 (/website/trunk/docs/en): Add QrCode link, image and instructions
or: ioerror committed revision 21801 (/website/trunk/en): Add android image and information to downloading page
or: ioerror committed revision 21802 (/website/trunk): add some specific android version tags
or: ioerror committed revision 21803 (/website/trunk/docs/en): Link to our packages directly
or: ioerror committed revision 21804 (/website/trunk/include): add signature tags
or: ioerror committed revision 21805 (/website/trunk/docs/en): add component build information
or: ioerror committed revision 21806 (/website/trunk/include): update component build information
or: ioerror committed revision 21807 (/website/trunk/en): tpyo
toruser
hi
where can i get info on all exit relays currently running and their status?
phobos
exitlist.torproject.org
toruser
tks
nsa
or: ioerror committed revision 21808 (/website/trunk/en): add direct links to versions
or: ioerror committed revision 21809 (/website/trunk/docs/en): add tag around stuff
or: phobos committed revision 21810 (/website/trunk/en): add the 0.2.1.24-stable announce item.
or: ioerror committed revision 21811 (/website/trunk/docs/en): fix broken link to packages
or: ioerror committed revision 21812 (/website/trunk/docs/en): add version number to webpage
or: ioerror committed revision 21813 (/website/trunk/en): aesthetic cleaning
outofwords
Do "A KeywordLine begins with a Keyword" means that any whitespace can happen before keyword? "\n signing-key\n" is valid?
Sebastian
I think that means that is invalid.
outofwords
so code is broken
Sebastian
also, please repaste all your good patches. I'll hang on to them now.
outofwords
patches do not helps. root is broken.
Sebastian
root?
outofwords
the spec can be parsed with any opinions. the canonical code do different than any pure java implementation.
you can close 1270.
xtoaster
anyone there ?
i got a situation
Sebastian
describe it
xtoaster
it seemed that much of the relays are blocked again :(
Anmibe
hi erverybody
Recently I installed my relay daemon on Mac OS X and up to now it works great, but there some issues..
1.) The paths on Mac OS X are strange, aren‘t they? 2.) I think it would be better to run tor in its own uid & gid, it would more secure
Sebastian
Which bundle did you install, and maybe you can give more details?
(For question 1)
For question 2: The expert bundle should run Tor as user tor
Anmibe
the recent expert paxkage from the website; well, my idea, ok something went wrong, no user/group tor; will have to look what I have done wrong
For me it would be more Mac like to find the files in /Library/Application Support/Tor/, including the files from user directory, instead of /Library/Tor & $Home (I know I can change it)
Sebastian
Feel free to file a bug if you care.
Anmibe
I installed the expert package 0.2.2.9-alpha for x86
nickm
outofwords, Sebastian: I am not averse to tightening the spec; I don't think that allowing space at the beginning of a line buys us anything, and if space _is_ allowed at the start of lines, our parsing code is incorrect. No Tor version generates documents with space at the start of lines, so forbidding it wouldn't hurt anything AFAICT.
outofwords
it's very hardly to do.
zBeeble
might have just had a DDOS attack against our TOR node.
nickm
Hard to allow space? Hard to forbid space?
outofwords
forbid. with my fixes if parsed line is "\n \n" thet returns "Unexpected EOF"
*the end of line
nickm
If the end of the line is "\n \n", that's an end of a line and a extra blank line that has only a space in it...
outofwords
yes. but we can't eat space before get_next_token()
nickm
Allowing space at the start of lines is hard too: it basically means a rewrite of router_get_hash_impl and everything that searches for "\nkeyword".
If space at the start of a line is disallowed, a line containing only space is disallowed.
Do those lines occur in practice? If not, banning them is fine IMO, unless I'm missing something
outofwords
shouldn't occur.
only if someone want to play with different chars for distinguish of clients, for rend stuff as example.
so right now such "\n \n" is ignores.
nickm
If we make _any_ change to parsing, it will be possible to tell old clients from new clients. That can't be helped.
outofwords
but most stuff filtered by auths, so it's less harms.
nickm
If we ban space at the start of a hidden service descriptor line, that's another way to tell old clients from new ones. I still that's okay if making the change fixes a worse bug.
outofwords
what about spaces after keywords without arguments?
nickm
Maybe router_get_hash_impl_helper() should make sure that the character immediately after end_str is a space of some kind.
I am less confident that those are safe to forbid; it is far easier to add them by mistake when coding.
("is a space of some kind" == "TOR_ISSPACE()")
outofwords
is "router-signature \n" allows by spec?
nickm
I am not sure.
outofwords
why any similiar KeywordLine with abother keywords could contain extra spaces?
"signing-key" NL means allowed in between?
nickm
You mean, why would that happen? "Programmer error" would be my guess.
outofwords
if "signing-key \n" disallowed so code of parser is wrong.
nickm
The way I read the spec, "signing-key" NL means space is _not_ allowed, but 1.2 implies that it _is_ allowed, the way I read it.
outofwords
the same here.
nickm
I think what we need to do is to figure out what the spec *should* say.
outofwords
yes.
nickm
I want it to say "no space at the start of the line. Space okay at the end of the line."
I think that allowing space at the end is important because sometimes we say stuff like sprintf(..."keyword %s\n",s) without making completely sure that s is nonempty.
outofwords
if I am correct most another implementations of parsers do simple "\nrouter-signature\n" no space.
if keyword allows arguments then spec says about it.
so sprintf(..."keyword %s\n",s) possible only for those keywords
nickm
Ah. So "trailing space is only allowed for keywords that can take arguments."
That would work.
outofwords
1.2 still makes two opinions about that statement.
nickm
Right; it should be clarified.
I am proposing "trailing space is only allowed for keyords that can take arguments" as a new rule, not as the only good reading of the existing rules.
outofwords
sounds good.
nickm
ok. I need to get groceries now; I will try to be online more later.
outofwords
"router-signature" NL Signature NL. Signatures is argument, no any keywords without arguments. "trailing space is only allowed in arguments" how is better?
it's object not argument in terms of code only.
and 1.2
or I am wrong.
RET_ERR("Unexpected EOF"); possible change it to RET_ERR("Unexpected EOF. Empty line with no keyword"); it's a little more informative to be.
ah yet for unrecognized keyword need to check it position from start of line.
« prev next »