1 2010-04-20T00:04:32 <ei-grad> ThomasWaldmann: i tested for all resonable values of ldap.OPT_X_TLS option — ldap.OPT_X_TLS_NEVER = 0, ldap.OPT_X_TLS_HARD = 1, ldap.OPT_X_TLS_DEMAND = 2, ldap.OPT_X_TLS_ALLOW = 3, ldap.OPT_X_TLS_TRY = 4... for server, which support start_tls extension, start_tls() raises ldap.SERVER_DOWN with ldap.OPT_X_TLS_HARD, and normally starts TLS with other values... now trying to find documentation for
2 2010-04-20T00:04:34 <ei-grad> them...
3 2010-04-20T00:06:41 <ThomasWaldmann> yeah, we shouldn't break stuff in 1.9 :)
4 2010-04-20T00:11:19 <ei-grad> seems like I should look into source code to make it clear how these values influence the work of python-ldap)
5 2010-04-20T00:14:51 <ei-grad> hm... or they may be documented in openldap manuals...
6 2010-04-20T00:16:17 * _buck says gn .
7 2010-04-20T00:16:24 *** _buck
8 2010-04-20T00:20:28 <ei-grad> heh) I was wrong twice) that values are for LDAP_OPT_X_TLS_REQUIRE_CERT )
9 2010-04-20T00:36:47 <ei-grad> OPT_X_TLS undocumented in python-ldap and openldap manuals... diving into openldap source)
10 2010-04-20T00:42:39 <dreimark> ei-grad: openldap or ldap is on freenode too
11 2010-04-20T00:43:40 * dreimark finished first version of barcamp press release
12 2010-04-20T00:43:57 <dreimark> gn
13 2010-04-20T00:46:07 <ei-grad> gn... /me going to bed too, tomorrow will ask about LDAP.ld_options.ldo_tls_mode on #openldap )
14 2010-04-20T01:46:33 *** AshishG
15 2010-04-20T04:06:26 *** mikema
16 2010-04-20T04:14:48 *** ValentinJaniaut_
17 2010-04-20T04:14:48 *** ValentinJaniaut
18 2010-04-20T04:14:49 *** ValentinJaniaut_
19 2010-04-20T04:58:21 *** mikema
20 2010-04-20T05:06:24 *** ValentinJaniaut_
21 2010-04-20T05:06:24 *** ValentinJaniaut
22 2010-04-20T05:06:25 *** ValentinJaniaut_
23 2010-04-20T05:16:16 *** kursor
24 2010-04-20T05:16:51 *** kursor
25 2010-04-20T05:17:10 *** kursor
26 2010-04-20T05:33:12 *** kursor
27 2010-04-20T07:24:58 *** _buck
28 2010-04-20T07:53:37 *** _buck
29 2010-04-20T10:00:52 <dreimark> moin
30 2010-04-20T10:01:02 <ei-grad> moin
31 2010-04-20T10:02:32 <ei-grad> dreimark: do you use jabber?
32 2010-04-20T10:04:34 *** DmitryAndreev
33 2010-04-20T10:06:10 <dreimark> not recently
34 2010-04-20T10:06:33 <dreimark> but i can try to reactivate it. irc is in the office for me much easier
35 2010-04-20T10:06:48 <dreimark> because i can go to moin shell by ssh and use irrsi therer
36 2010-04-20T10:06:57 <dreimark> so no firewall issues
37 2010-04-20T10:09:11 <ThomasWaldmann> moin
38 2010-04-20T10:10:25 <ThomasWaldmann> irc is preferred way of "moin" communication, because this channel is shared by the devs and even logged, so one can lookup stuff later
39 2010-04-20T10:24:14 <ei-grad> i asked openldap developers about LDAP_OPT_X_TLS, http://pastebin.ca/1869249 , so it seems like there is no need to set it at all... and only behaviour that it has, is that if it is set to OPT_X_TLS_HARD, then start_tls is called automatically in ldap_open... so LDAPAuth with start_tls=1 calls ldap_start_tls twice, and gets an error... that is why I can't to connect to server that don't support
40 2010-04-20T10:24:16 <ei-grad> start_tls, when passed start_tls=1 in LDAPAuth (in moin source it is called as "to try TLS", and really it is OPT_X_TLS_HARD)
41 2010-04-20T10:24:22 *** DmitryAndreev
42 2010-04-20T10:30:15 <ThomasWaldmann> ei-grad: should be implemented in a way that "i want only tls" is possible. in that case, it shouldn't do without tls if tls doesn't work, but complain.
43 2010-04-20T10:37:02 *** DmitryAndreev
44 2010-04-20T10:51:45 <waldi> my usual answer: starttls should be done whenever supported by the server even if no means of validating the identity exists. tls themself provides perfect forward security and resistance against passive attacks. tls supports certificate-less ciphers for this
45 2010-04-20T10:56:33 *** ValentinJaniaut
46 2010-04-20T11:01:25 <ei-grad> ok, I understand it
47 2010-04-20T11:01:27 <ei-grad> bbl
48 2010-04-20T11:01:45 <ThomasWaldmann> btw, students who like to know detailed scoring feedback about task2 just grab me or TheSheep and ask
49 2010-04-20T11:03:01 *** ValentinJaniaut
50 2010-04-20T11:24:18 *** gwork
51 2010-04-20T12:49:53 <ei-grad> dreimark: ThomasWaldmann: http://dpaste.com/hold/185679/ ... think it can be included in 1.9...
52 2010-04-20T13:50:06 *** mikema
53 2010-04-20T14:05:07 *** _buck
54 2010-04-20T14:21:25 *** DmitryAndreev
55 2010-04-20T14:35:33 <_buck> moin
56 2010-04-20T14:36:37 <_buck> Which date are we expecting the allocation results, 21st or 26th ?
57 2010-04-20T14:38:02 <dreimark> ei-grad: please add a comment self.start_tls == 2 for the number at that point
58 2010-04-20T14:38:09 <dreimark> waht it is
59 2010-04-20T14:38:31 <dreimark> _buck: when it is official we will tell, we don't want to make noise
60 2010-04-20T14:40:45 <_buck> ok no google has written that on 21st lists will be made and on rest of the days till 26th the collisions will be addressed. I am just asking which date will we know about results?
61 2010-04-20T14:41:05 <xorAxAx> 26th
62 2010-04-20T14:41:14 <_buck> oh k
63 2010-04-20T14:41:36 *** diofeher
64 2010-04-20T14:44:52 <diofeher> moin moin
65 2010-04-20T14:45:32 <diofeher> i was looking at logs, and saw that if i want to see score just ask to ThomasWaldmann or TheSheep
66 2010-04-20T14:47:29 <TheSheep> scroing feedback, not score :)
67 2010-04-20T14:47:38 <TheSheep> scoring*
68 2010-04-20T14:49:12 <diofeher> ah ok
69 2010-04-20T14:49:15 <diofeher> >)
70 2010-04-20T14:49:22 * dreimark weird text
71 2010-04-20T15:40:23 *** _buck
72 2010-04-20T15:45:04 *** DmitryAndreev
73 2010-04-20T15:45:11 *** DmitryAndreev
74 2010-04-20T15:46:44 *** diofeher_
75 2010-04-20T15:48:13 *** diofeher
76 2010-04-20T15:48:25 *** diofeher_
77 2010-04-20T16:25:03 *** gwork
78 2010-04-20T16:48:35 *** diofeher
79 2010-04-20T17:16:29 *** _buck
80 2010-04-20T18:02:22 <ValentinJaniaut> TheSheep: ThomasWaldmann, I would be glad to have any feedback on task2 when you have time
81 2010-04-20T18:17:33 *** JosefMeier
82 2010-04-20T18:18:07 <JosefMeier> Does someone know, in which file and directory the page cache is in Moin 1.9.2 ??? Can't find it.
83 2010-04-20T18:18:18 <JosefMeier> Getting the list of pages takes very long. Maybe I don't have an activated cache for some reason.
84 2010-04-20T18:18:43 <JosefMeier> Or is page title caching performed by XAPIAN?
85 2010-04-20T18:21:45 <dreimark> do you have looked into the macro?
86 2010-04-20T18:22:12 <JosefMeier> in which one?
87 2010-04-20T18:22:22 <dreimark> for titleindex
88 2010-04-20T18:22:35 <JosefMeier> Is also xmlrpc using this macro?
89 2010-04-20T18:22:48 <JosefMeier> for getAllPages
90 2010-04-20T18:22:58 <dreimark> then you need to look in this
91 2010-04-20T18:23:22 <ei-grad> dreimark: http://dpaste.com/hold/185806/ a comment like that?
92 2010-04-20T18:23:31 <dreimark> JosefMeier: http://hg.moinmo.in/moin/1.9/file/c7dd42405e5a/MoinMoin/xmlrpc/__init__.py#l264
93 2010-04-20T18:24:07 <dreimark> ei-grad: yes
94 2010-04-20T18:24:37 <JosefMeier> dreimark: I'm just confused cause I don't find a cache file with page names in it. Do you know where the cache file for page titles is? Do you use a different algorithm in 1.9.2 compared to 1.8.x ?
95 2010-04-20T18:26:18 <dreimark> JosefMeier: there is none
96 2010-04-20T18:26:33 <dreimark> just look at the line 318 pagelist = self.request.rootpage.getPageList
97 2010-04-20T18:26:36 <dreimark> ...
98 2010-04-20T18:27:47 <dreimark> you get a faster result if you use xmlrpc_getAllPagesEx and add options to exclude underlay
99 2010-04-20T18:28:37 <JosefMeier> Hm. What's the reason why there's no page title cache?
100 2010-04-20T18:30:00 <dreimark> how should it work?
101 2010-04-20T18:31:33 <JosefMeier> Like every other write through cache in Moin. I mean: is there a good reason for not having a page title cache?
102 2010-04-20T18:32:41 <JosefMeier> I only have around 300 pages in my wiki and getting all page titles (without system and underlay pages) takes a few seconds.
103 2010-04-20T18:35:25 <dreimark> http://hg.moinmo.in/moin/1.9/file/c7dd42405e5a/MoinMoin/Page.py#l1726
104 2010-04-20T18:35:39 <dreimark> JosefMeier: see there and verify if you don't have this cache
105 2010-04-20T18:36:13 <dreimark> btw. for which platform do you see these problems
106 2010-04-20T18:36:50 <JosefMeier> on WinXP
107 2010-04-20T18:37:01 <JosefMeier> with apache modwsgi
108 2010-04-20T18:37:06 *** mikema
109 2010-04-20T18:37:15 <dreimark> and which broken filesystem ?
110 2010-04-20T18:37:26 <JosefMeier> I like your humor.
111 2010-04-20T18:37:51 <JosefMeier> dreimark: ntfs
112 2010-04-20T18:38:11 <JosefMeier> "broken filesystem" *ggg*
113 2010-04-20T18:38:16 <xorAxAx> ThomasWaldmann, waldi: who of you will be in the deduplication meeting?
114 2010-04-20T18:38:33 <dreimark> he, it is because of the problems we run into it for valiating by time
115 2010-04-20T18:38:39 <dreimark> +d
116 2010-04-20T18:40:42 <waldi> this is tomorrow
117 2010-04-20T18:41:49 * dreimark we have nice students
118 2010-04-20T18:51:52 <dreimark> xorAxAx: reload that page
119 2010-04-20T18:53:51 <xorAxAx> which page?
120 2010-04-20T18:55:15 <dreimark> list proposal
121 2010-04-20T18:57:24 <waldi> yeah, no list any longer
122 2010-04-20T19:03:21 <JosefMeier> dreimark: my questions are answered: page title's are cached IN MEMORY and the big delay comes from the code with is filtering out underlay and system pages + acl checks.
123 2010-04-20T19:03:39 <JosefMeier> s/with/which/
124 2010-04-20T19:05:51 <dreimark> JosefMeier: ok, so excluding underlay from the beginning would make it faster.
125 2010-04-20T19:06:48 <JosefMeier> dreimark: But for some reason getPageList for 1.8.x is much faster on my machine as for 1.9.2
126 2010-04-20T19:07:03 <JosefMeier> I'll try to find the reason...
127 2010-04-20T19:07:11 <dreimark> how much underlay pages do you have
128 2010-04-20T19:07:24 <dreimark> is it only german or the complete hg set
129 2010-04-20T19:08:35 <JosefMeier> only the german ones
130 2010-04-20T19:09:25 <JosefMeier> 91 pages are in the underlay dir
131 2010-04-20T19:09:30 <dreimark> strange
132 2010-04-20T19:13:07 <JosefMeier> I'll doing some debugging for that and give you a note if I know more.
133 2010-04-20T19:15:02 <ThomasWaldmann> re
134 2010-04-20T19:22:33 <dreimark> hmm, we convert a name of V.S.O.P..png to V.S.O.P_png
135 2010-04-20T19:32:12 <dreimark> that creates a mimetype problem
136 2010-04-20T19:52:57 <ThomasWaldmann> use one dot
137 2010-04-20T20:00:44 <dreimark> it happens by xmlrpc uploading files
138 2010-04-20T20:04:48 *** JosefMeier
139 2010-04-20T20:05:52 <ThomasWaldmann> it's likely some sanitizing function going on red alert when it sees ..
140 2010-04-20T20:08:22 *** moin2
141 2010-04-20T20:08:29 *** moin2
142 2010-04-20T20:09:10 *** JosefMeier
143 2010-04-20T20:16:06 *** kursor
144 2010-04-20T20:18:07 *** kursor
145 2010-04-20T20:22:28 *** _buck
146 2010-04-20T21:01:49 <ThomasWaldmann> mentor meeting in 1h
147 2010-04-20T21:05:04 * dreimark ok
148 2010-04-20T21:58:38 *** JosefMeier
149 2010-04-20T22:27:43 <ThomasWaldmann> xorAxAx: TheSheep: waldi: you join us on the other channel?
