IRC log for #brlcad on 20130129

00:22.28*** join/#brlcad KimK (~Kim__@wsip-184-176-200-171.ks.ks.cox.net)
00:34.41*** join/#brlcad dtidrow (~dtidrow@c-68-84-167-135.hsd1.mi.comcast.net)
00:52.02*** join/#brlcad KimK (~Kim__@wsip-184-176-200-171.ks.ks.cox.net)
01:14.56*** join/#brlcad Notify (~notify@66-118-151-70.static.sagonet.net)
01:23.03*** join/#brlcad KimK (~Kim__@wsip-184-176-200-171.ks.ks.cox.net)
01:52.39*** join/#brlcad KimK (~Kim__@wsip-184-176-200-171.ks.ks.cox.net)
02:23.05*** join/#brlcad KimK (~Kim__@wsip-184-176-200-171.ks.ks.cox.net)
03:15.06Notify03BRL-CAD Wiki:Carpinteyrorcz * 0 /wiki/User:Carpinteyrorcz:
03:40.48*** join/#brlcad agkphysics (~aaron@206.10.252.27.dyn.cust.vf.net.nz)
03:54.31*** join/#brlcad aaron (~aaron@206.10.252.27.dyn.cust.vf.net.nz)
04:11.30*** join/#brlcad EricPoehlsen (~Eric@dslb-084-056-082-235.pools.arcor-ip.net)
04:13.46*** join/#brlcad Guest24238 (~aaron@206.10.252.27.dyn.cust.vf.net.nz)
05:43.41Notify03BRL-CAD:brlcad * 54210 (brlcad/trunk/src/conv/step/DerivedUnitElement.cpp brlcad/trunk/src/conv/step/DerivedUnitElement.h and 2 others): fixed bad header. looks like 'United' regex gone mad.
05:44.30brlcadstarseeker: your src/librt/test_* files seem to be lacking their header (sh/header.sh will fix it for you)
06:12.29Notify03BRL-CAD:brlcad * 54211 brlcad/trunk/src/tclscripts/mged/botedit.tcl: add missing header and footer
06:14.29Notify03BRL-CAD:brlcad * 54212 brlcad/trunk/sh/template.sh: if /bin/sh isn't a bash shell, footer.sh will fail (due to the array usages). make sure we're using bash. clean up backup recovery outputs too.
06:19.29Notify03BRL-CAD:brlcad * 54213 (brlcad/trunk/src/tclscripts/boteditor/botEditor.tcl brlcad/trunk/src/tclscripts/boteditor/botPropertyBox.tcl brlcad/trunk/src/tclscripts/boteditor/botTools.tcl): add missing headers and footers
06:23.44Notify03BRL-CAD:brlcad * 54214 brlcad/trunk/src/tclscripts/archer/itk_redefines.tcl: missing header/footer added
06:27.11Notify03BRL-CAD:brlcad * 54215 brlcad/trunk/src/libsysv/tcl.c: add missing header, remove per-file authorship per dev guide
06:29.40Notify03BRL-CAD:brlcad * 54216 brlcad/trunk/sh/copyright.sh: ignore cmake build directories and svn dirs.
06:43.08*** join/#brlcad EricPoehlsen (~Eric@ip-109-84-173-38.web.vodafone.de)
07:22.22Notify03BRL-CAD:brlcad * 54217 (brlcad/trunk/CMakeLists.txt brlcad/trunk/COPYING and 2735 others): update copyright for 2013 welcoming our files to the year of slithering slathering serpent subjects
07:48.45Notify03BRL-CAD Wiki:033-compare-online-loans-no-credit-check-approval * 0 /wiki/User:033-compare-online-loans-no-credit-check-approval:
07:56.14*** join/#brlcad d_rossberg (~rossberg@66-118-151-70.static.sagonet.net)
13:15.35*** join/#brlcad EricPoehlsen (~Eric@ip-109-85-12-50.web.vodafone.de)
15:29.36brlcadtightens up mediawiki to not allow dashes in username and usernames > 12char
15:30.11brlcadstill don't know why the MinimumNameLength extension isn't catching those
15:30.27brlcadmodified the core User.php file to impose harder limits
15:31.01Notify03BRL-CAD Wiki:Jacksixb * 4942 /wiki/User:Jacksixb/Independent_Study_2012-13_Log:
15:33.54brlcadstarseeker: any hints on how I can fix this? http://pastebin.ca/2308596
15:34.44brlcadat least what the right fix is, clearly caused by not finding fontconfig package, but is the fix to empty out the var so it's not NOTFOUND?
15:36.42maths22brlcad: why did you want to do that?
15:38.02maths22lastlog -
15:38.06maths22sorry
15:42.46maths22it ist to prevent spam users?
15:56.57brlcadyep
15:58.03brlcadmaths22: I installed an extension to limit usernames, but it doesn't seem to prevent them fully
15:58.13brlcadnote the recent..
15:58.13brlcad02:48 < Notify> BRL-CAD Wiki:033-compare-online-loans-no-credit-check-approval * 0 /wiki/User:033-compare-online-loans-no-credit-check-approval
15:59.29brlcadwe're also still getting 1-2 spams a day, so I'm looking for more measures that don't impede fly-by-editing
15:59.44brlcadour regex is getting pretty huge
16:04.48maths22true.
16:05.42brlcadwe had it down to about 1-2 a month, which was quite acceptable
16:06.00brlcadbut then hundreds of spam accounts were getting created, nice to limit that too
16:06.35brlcadan option to double-verify new accounts would be nice..
16:08.32``Erikdisable new page creation for non-verified users?
16:09.38maths22what do you mean double-verify
16:11.02``Erika common approach is to require an email address and send an email there with a validation link to 'enable' an account, would that be a useful approach?
16:11.23maths22Let me look into that.
16:11.26brlcadusers must already provide an e-mail address I "think"
16:11.38brlcadhaven't tried to create a new account in a while.. ;)
16:12.13brlcadrequiring an account for new pages would be good if that's not already enforced
16:12.26brlcadstill want to allow anonymous edits to any page
16:13.15maths22by they half to confim it.
16:13.19maths22Let me look
16:13.50``ErikI'd imagine some of the spam is generated mechanical turk style :/
16:14.43brlcadthe majority are rather complex but at least half-automated
16:15.09brlcadthere was one that clearly got posted in error a while back that showed the regexes the spam script was using
16:15.25brlcadit had it's own internal regexes to get past filtering
16:16.05``Erikcool, didja save it anywhere? be interesting to see the red teams secret technologies in this arms race
16:16.18brlcad(buy|purchse|sell) (car|boat|house) (loans|investments|estate) on the best (property|deal|goal) ...
16:16.42brlcadi'm sure it's in the wiki history somewhere, but no not specifically
16:16.55brlcadjust gave me a couple dozen patterns to add to our regex ;)
16:17.13maths22I have deactivated page creation for unregistered users
16:17.21brlcadmaths22: awesome
16:18.06brlcadwhat was REALLY effective was disabling all URLs ...
16:18.16maths22what do you mean.
16:18.22brlcadbut of course then we couldn't even edit our own existing pages
16:18.37brlcadthe regex pattern match extension
16:18.44brlcadI modified it to match any url
16:18.50``Erikif text contains "http://" reject
16:18.58brlcadlil more complex, but yeah
16:19.16brlcad"http:\/\/.*\.com|http:\/\/.*\.net|http:\/\/.*[^cC][^aA][^dD]\.org|". # any link to an external site
16:19.46maths22that would make using our own site hard.
16:19.48brlcadbut that was a no-go, way too restrictive
16:20.19brlcadwhat WOULD be useful is if a submit contains a url, that it prompts an additional captcha
16:20.33brlcadeven better, if it contains a *new* url :)
16:20.38``Erik'cept it was fine for the mountains of spam to http://mrastercard.com.cn *cough* :)
16:20.42brlcadbut I'm not aware of such an extension
16:21.19``Erik.cn is china
16:21.25brlcad``Erik: that'd still get caught by http:\/\/.*\.com
16:21.31``Erikah, hm
16:21.48brlcadbut yeah, some isoteric matches *could* get through, just unlikely :)
16:22.06``Erikregisters brlcad.xxx before it becomes a hot property O.o
16:22.09brlcadmyawesomespamcad.org would get in
16:22.48brlcadand any other tld actually not containing ".com", ".net", or "cad.org"
16:23.12brlcadthey're just currently not even in the noise
16:24.24maths22look at this: http://www.mediawiki.org/wiki/Extension:QuestyCaptcha
16:26.42brlcadmaths22: we already have that installed
16:26.50brlcadvia ConfirmEdit
16:27.18brlcadI think that captcha is only enabled during an edit
16:27.29brlcadpretty effective at edit spam
16:28.25brlcadhell, most fly by users would have to do some research to answer the questions I put in there... :)
16:28.30brlcade.g., What is the dot product of [2, 12, 1] with [7, 2, 4]
16:30.40maths22Why are we allowing all registered users to skip the captcha
16:30.54maths22We should only let autoconfirmed users skip it.
16:31.00brlcadbecause I don't want to burden a real user with a real edit
16:31.17brlcadthere's that distinction?
16:31.20maths22what about this: $wgGroupPermissions['emailconfirmed']['skipcaptcha'] = true;
16:31.39brlcads/there/where/
16:32.07brlcadthat sounds good to me, I assumed you couldn't be a registered user without confirming your e-mail
16:32.10maths22Sorry.  Autoconfirmed means:
16:32.31maths22never mind.
16:32.40maths22What we should do is set http://www.mediawiki.org/wiki/Manual:$wgAutoConfirmAge
16:32.53maths22and/or http://www.mediawiki.org/wiki/Manual:$wgAutoConfirmCount
16:34.36maths22or rather http://www.mediawiki.org/wiki/Manual:$wgAutoPromote
16:34.59brlcadautoconfirmage doesn't look very useful
16:35.19brlcadmuch of the spam accounts are registered several days before they start spamming
16:35.30brlcadand they are then only used to spam once or twice
16:35.52brlcadprobably to combat exactly that
16:36.13maths22then use autoconfirmcount
16:36.32brlcadyeah, that sounds useful
16:36.34maths22ignore my last link
16:36.38brlcadwhat's a good value?
16:36.40brlcad3?
16:37.03brlcadeffectively 0 now right?
16:37.23maths22yes.
16:37.29maths223 to five sounds good
16:37.44brlcadyeah, a few captchas when they're just getting started sounds reasonable
16:37.52maths22ok.
16:38.32brlcadmore than 3 is probably not preventing anything -- if they get past even one, it's a real human and they'll be able to get past N
16:40.34maths22true
16:40.47maths22I have set that
16:42.00brlcadwhat about captcha during account creation?
16:42.10brlcaddoes it currently prompt?
16:42.50maths22It already does
16:44.44brlcadbut not the questy one, I imagine?
16:45.56maths22yes. it is the questy one
16:46.04brlcadhuh
16:46.34maths22I remember I had to think about the sum of elements in a zero matrix.
16:46.41brlcadunrelated, any way to return users that log in from drupal to the drupal page they logged in from?
16:46.54maths22Let me look at that
16:47.53brlcadif you all have other good CADy mathy matrixy questions to include, lemme know ;)
16:58.11``Erik<-- kinda wonders if a dot product isn't too much, plenty of people do modelling and purdee pikchurz without doing that much math regularly O.o
16:58.58``Erikbasic 4 operations might be good, I've seen those used as captchas before
17:02.06brlcad``Erik: yet a simple search reveals how to compute a dot product with trivial ease
17:02.13brlcadeven online calculators
17:03.03brlcadI don't think we have a single user that couldn't figure it out if they had to
17:03.14``Erikyeah, it's easy to do, but I'd guess that a lot of non-geeks wouldn't think to google it *shrug*
17:03.20brlcadand they get a different captcha if they fail anyways
17:03.59``Erik(not that the contribution rate is high among new/driveby users *shrug*)
17:04.38brlcadit's non-zero, so that's in part why I still want to support anonymous editing at all
17:04.42brlcadI think it works
17:04.58brlcadwe've had a half dozens anon edits in the past two months
17:05.25``Erikaight, be interesting to collect the data from the old machine on contribution rates and compare it to that kind of captcha
17:05.43``Erikd'no if there's enough to do a statistically significant a/b test :/
17:06.01brlcadquestions in our domain that require thought are okay by me, just can't be "hard" questions (as in not easily searched or understood)
17:07.01``Eriksooo, not "provide a ponynomial time algorithm to the traveling salesman problem in haskell"
17:09.20brlcadyeah
17:09.26brlcadhaskell would be too much ;)
17:13.58maths22let's see if autoconfirm + no anonymous pages helps
17:14.27brlcadthat's no anonymous NEW pages right?
17:14.35brlcadstill want to allow anonymous edits on existing pages
17:14.45``Erikhow's this for mental sickness: http://paste.lisp.org/display/134995 (sh script for cron)
17:16.15brlcadwhat for?
17:16.57``Erikin case data accidently gets stomped
17:17.14brlcadokie
17:18.05brlcadI like to keep a local and offsite backup of dumps, have it scripted up for the mysql dbs
17:18.17brlcadwas what was filling up .bz's hard disks a couple years ago if you remember
17:19.08``Erikyeah, I had a nightly script on forge that did db dumps before an fs dump, good to have a replayable script
17:51.12*** join/#brlcad EricPoehlsen (~Eric@dslb-084-056-082-235.pools.arcor-ip.net)
18:32.34brlcadpokes Notify
18:59.38brlcadhttp://pastebin.ca/2308674
19:02.50``Eriklast email notify got was at 11:48:15 and was posted O.o
19:03.12``Erikvar is fun
19:03.14``Erikfull
19:05.34Notify03BRL-CAD:tbrowder2 * 54221 (brlcad/trunk/doc/docbook/resources/brlcad/images/logo-vm-gears.png brlcad/trunk/doc/docbook/resources/brlcad/images/logo-vm-gears.svg): change to 128x128 per web site
19:11.07brlcadahh, /var/log is growing now that we're all on it
19:11.24brlcadheh, web logs
19:11.36brlcadalready 3GB
19:12.26``Erikmebbe move 'em to /backup or /usr
19:12.49``Erikd'no what the plans are for /backup, but it's a different spindle
19:13.22brlcadyeah. no plans
19:13.45brlcadit came that way, we can make it be anything but a level 0 backup of the main disk is actually kinda useful
19:14.09brlcadof course, that's half a tb, which we could use if space gets tight
19:14.41brlcadprobably just move var/log to usr/var/log like the other dirs for now
19:14.50``Erikghetto raid-1 done with dd or something?
19:14.50brlcads/move/link/
19:14.55brlcadyeah
19:15.03brlcadrsync it
19:15.06``Erikor cpio once it's got the first image
19:15.36brlcadsync it weekly with no deletes or similar
19:18.12``Erikthe notify posts should show up over the next few hours in random order as smtp retries come in
19:19.00``Erikassuming the 47 free megs don't get used up first O.o
19:22.17``Erikturns off the hourly tarball regeneration of the git mirror.. should slow down the logs a bit
19:32.15Notify03BRL-CAD:r_weiss * 54222 brlcad/trunk/src/tclscripts/mged/lc.tcl: Update to the mged "lc" command adding the "-r" option to skip listing regions which their parent is a region and the child is subtracted within the parent. Also added the 'los' test to the "-s" option.
19:32.29brlcadlets see how hard this hits the fan
19:34.15brlcadwaits on apachectl stop && mv /var/log /usr/var/log && ln -s /usr/var/log && apachectl start
19:36.58brlcaddone, now hopefully the mail spool will be a lil more resiliant
19:36.58``Erikall of /var/log, not just /var/log/httpd ?
19:37.08brlcadwhole deal
19:37.16brlcaduser logs are huge once they get re-enabled
19:37.50``Erikcould cause boot issues :/
19:38.07brlcadhrm, how so?  shouldn't care..
19:38.42brlcadI've had /var/log redirected elsewhere without a problem iirc
19:39.27``Erikah, syslogd isn't started until thhe FILESYSTEMS and criticalremotes
19:39.31``Erikaight, cool
19:40.06brlcadhups syslogd just for good measure
19:41.27starseekerbrlcad: I think the fontconfig fix looks OK
19:41.39``Eriklsof seems happy
19:42.11starseekermutters under his breath some more about Tcl/Tk...
19:43.53brlcadmaths22: do you recall when you submitted your first task (or first popped on irc)?
19:45.43brlcadi'm seeing nov I think
19:47.15``Erikheh, var is now 3% used
19:48.31Notify03BRL-CAD:brlcad * 54218 brlcad/trunk/src/other/tk/CMakeLists.txt: also need to account for a non-existent fontconfig
19:48.51Notify03BRL-CAD Wiki:Lajosgodly * 0 /wiki/User:Lajosgodly:
19:55.26Notify03BRL-CAD:brlcad * 54223 brlcad/trunk/AUTHORS: credit burroughs (maths22) for his awesome work getting most of our website infrastructure migrated to newer hardware and fixing/improving numerous website features (upgrades, search, unified login, spam filtering, more). came in under GCI but interested in taking over website ops.
20:03.31Notify03BRL-CAD:n_reed * 54224 (brlcad/trunk/src/librt/librt_private.h brlcad/trunk/src/librt/primitives/pipe/pipe.c brlcad/trunk/src/librt/primitives/primitive_util.c): replace pipe draw routine struct parameters & returns with pointer parameters to avoid struct copies
20:14.00Notify03BRL-CAD:brlcad * 54225 brlcad/trunk/src/other/openNURBS/opennurbs_array.h: gcc 4.4.6 warns about being unable to inline the (implicit) destructor. declaring the destructor implies to not inline, making the warnings/errors go away. performance should be moot, destructor is ON_SimpleArray's virtual.
20:25.08Notify03BRL-CAD:tbrowder2 * 54220 (brlcad/trunk/doc/docbook/resources/brlcad/images/logo-vm.svg =================================================================== and 12 others): rename file to match that used on the site
20:25.17*** join/#brlcad luca79 (~luca@net-2-35-211-123.cust.dsl.teletu.it)
20:28.06Notify03BRL-CAD:brlcad * 54219 (brlcad/trunk/src/other/boost/boost/detail/container_fwd.hpp brlcad/trunk/src/other/boost/boost/mpl/has_xxx.hpp): apply boost patch regarding error: "BOOST_MPL_HAS_XXX_NO_EXPLICIT_TEST_FUNCTION" is not defined along with a handful of other defines not defined. see https://svn.boost.org/trac/boost/ticket/6142 for some relevance.
20:35.27Notify03BRL-CAD:brlcad * 54226 brlcad/trunk/src/other/openNURBS/opennurbs_layer.h: more inline warnings/errors due to the compiler attempting to inline default functions. declaring them quells via intent.
20:40.13maths22brlcad: November 27
20:40.35maths22day after GCI started
20:41.05maths22yes.  No autonomous new pages
20:50.24Notify03BRL-CAD:carlmoore * 54227 brlcad/trunk/src/shapes/coil.c: simplify 2 more derivations of coil_radius
21:07.06Notify03BRL-CAD:brlcad * 54228 (brlcad/trunk/src/other/openNURBS/opennurbs_array.h brlcad/trunk/src/other/openNURBS/opennurbs_layer.h): revert 54226 and 54225 since they cause vtable symbol failures when linking against the lib (declaring them apparently also implied don't define the default.
21:11.53*** join/#brlcad luca79 (~luca@net-2-35-211-123.cust.dsl.teletu.it)
21:19.53Notify03BRL-CAD:carlmoore * 54229 brlcad/trunk/src/shapes/coil.c: remove commented-off code because of the simplicity of the replacement
21:30.33Notify03BRL-CAD:carlmoore * 54230 brlcad/trunk/src/shapes/coil.c: add decimal points although this should already compute in floating point mode
21:36.54Notify03BRL-CAD:carlmoore * 54231 brlcad/trunk/src/conv/iges/extrudcon.c: simplify computation of q1
22:39.29Notify03BRL-CAD:carlmoore * 54232 brlcad/trunk/src/conv/shp/shapelib/shpopen.c: simplify what I take to be integer-mode calculations

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.