[00:05] lifeless: i'm telling you, it's a liberal lie [00:16] SamB: bzr-svn is already using platform-specific auth providers where possible [00:17] jelmer: all of the ones that that function returns? [00:17] 'cause I didn't see bindings for it [00:18] there's one for windows and one for keychain (Mac OS X) [00:18] kwallet is missing [00:18] support for gnome-keyring is also not there, but already provided through bzr-gtk [00:26] SamB: are you still convinced this is a valid bug report? [00:35] jelmer: well, I do think it would make more sense to use that function than to do everything piecemeal ... though APR is a bit of a pain ... [00:36] and why is the gnome-keyring bit part of bzr-gtk? [00:36] svn uses it fine from the CLI ... [00:37] SamB: it's part of bzr-gtk because it's not svn-specific in any way [00:37] SamB: "plain" bzr can use gnome keyring as well [00:38] jelmer: oh. well, svn uses it in a particular way [00:38] SamB: that function doesn't really allow us to do everything piecemeal, we still have to call it with the name of the client provider we'd like to obtain [00:38] that's hardly no more useful than seeing if a method with a particular name exists in Python [00:38] s/no/any/ [00:39] jelmer: oh, wait, I must have got mixed up [00:39] SamB: related to this, I don't want to know what happens if both bzr-gtk and bzr-svn both register providers that use gnome keyring [00:41] SamB: what's the actual problem you're trying to solve? "must use function 'foo'" doesn't seem like a worthwile goal by itself [00:47] hi jelmer [00:47] moin poolie [00:48] jelmer: er, sorry about that, my computer has been rebooting at the darnedest times lately ... [00:48] jelmer: oh, wait, I must have got mixed up [00:48] that's the last thing I saw ... [00:48] SamB: related to this, I don't want to know what happens if both bzr-gtk and bzr-svn both register providers that use gnome keyring [00:48] SamB: what's the actual problem you're trying to solve? "must use function 'foo'" doesn't seem like a worthwile goal by itself [00:49] jelmer: anyway, right before it rebooted I realized you were probably looking at the singular-named function that is named almost the same as a plural-named function [00:50] * jelmer has another look [00:51] ah [00:51] SamB: you're right, that would probably be useful to expose on a subvertpy level [00:51] SamB: I don't see what problem it would solve for bzr-svn though [00:53] well, it would mean not having to do anything to support any new platform-specific authentication providers that SVN 1.7 might have ... [00:54] SamB: it also means duplication of gnome-keyring [00:55] SamB: any specific stuff that svn 1.7 might add we can add support for manually /if it doesn't conflict with stuff bzr already provides/ [00:57] jelmer: I don't understand this whole "conflict" thinking [00:59] SamB: if e.g. the password in gnome-keyring is wrong we'll end up sending it twice === fta_ is now known as fta [01:00] jelmer: say, is this related to seahorse at all? [01:00] because I don't *have* that [01:01] SamB: no, seahorse is only accessed by bzr-gtk for gpg signatures [01:01] 'kay [01:01] how do I get at what gnome-keyring stores? [01:02] I think there's a separate app [01:03] gnome-keyring-manager or something IIRC [01:04] jelmer: ah, that's it alright [01:06] jelmer: so, how is the bzr-gtk support supposed to work? [01:07] SamB: it registers a fallback credentials provider [01:07] SamB: that gets queried by bzr before it prompts [01:08] wonder why I didn't see anything like that happen? [01:09] ... that not need to ask me for a password before it could do anything? [01:09] +would [01:09] SamB: it'll only use existing passwords, never storing new ones [01:10] (since it doesn't have a way to tell whether a password that was entered and returned was valid [01:10] it can check whether there is one without the keyring password being entered? [01:11] I don't know [01:12] i've upgraded to a recent enough bzr and svn to set revision properties instead of file properties on push. can I delete all the bzr: file properties in svn without deleterious results? [01:13] dash: freenode appears to have eaten your "jelmer: " [01:24] jelmer: hmm, /usr/share/doc/gnome-keyring/keyring-intro.txt.gz tells me that it only asks for a password if something is found [01:24] jam: ping? Are you still looking at the inventory-delta patch? [01:25] so I'm thinking that either this isn't working at all for bzr, or bzr and svn don't find the same entries [01:26] dash: yeah, you should be able to do that. the main thing you'll probably lose is the ability to branch old revisions by revid easily [01:27] not an issue, i expect [01:29] dash: emphasis on *should*, this isn't something that a lot of people have been testing afaik [01:30] jelmer: Hooray! Science! [01:30] I will let you know if I develop any adverse symptoms. [01:30] SamB: yeah, that sounds sensible [01:31] Best way to fabricate commits? XD [01:34] jelmer: FWIW, the entry that SVN made in my keyring has only the Username, Password, Port, and Domain fields filled in in gnome-keyring-manager ... [01:34] and the Port field has a 0 in it [01:35] and is grayed out [01:35] so maybe that's not really filled in either [01:38] jelmer: how is the bzr user expected to have gotten entries in there that bzr would use ? [01:53] SamB: epiphany [01:53] SamB: also, hopefully the API will start supporting the addition of new entries at some point [01:54] jelmer: the bzrlib API? [01:56] jelmer: anyway ... it sure looks like your gnome-keyring thingy doesn't subsume SVN's ... [02:00] lifeless: want to review my inventory-delta patch? [02:00] spiv_: yes [02:00] I've been battling a check patch [02:00] hopefully its really sorted now, and I'm -> food [02:00] but after that [02:01] lifeless: hooray! Thanks :) [02:01] also evolution is being utterly retarded [02:01] lifeless: you ... don't evolve, though [02:01] you're an individual [02:01] this is *not* pokemon [02:01] I mean really. Its like a kid thats got diarrea crawling backwards and eating off the floor, all at the same time. [02:02] SamB: Dunno about you, but my cells divide, have random mutations and undergo a fitness test. [02:09] ew [02:09] enough, hey [02:10] sorry [02:11] 3 hours now I've been trying to get email read [02:11] I'm venting elsewhere now === spiv_ is now known as spiv [02:47] * emmajane blinks at lifeless. [02:48] * emmajane thinks that sometimes it's better not to read backwards. :) [02:56] spiv: I sent in a little bit more review, but I haven't "finished" yet [02:57] However, for *me* InventoryDelta is significantly *faster* for converting than IDS for bzr.dev @2000 [02:57] I think because the extra repacks are costing it a bit of time [02:59] jam: yeah, I see that too. [02:59] note that it is still *slower* for mysql-525 [03:00] I haven't split out the repacking time to know where the problem might be. [03:00] It might be repacking [03:00] I'm just in the process of grabbing launchpad so I have a larger dataset to work with. [03:00] repacks tend to crawl for me ... [03:00] it might also be that by default btreebuilder will cache up to 100,000 nodes before writing anything to disk [03:00] while repacking is causing it to write and read again much more often [03:01] spiv: except lp starts as --2a ... [03:01] might want to grab mysql instead [03:01] Or as well :) [03:03] spiv: so I wonder if the last bit is that IDS will not re-read an inventory from the source if it misses from the cache. [03:03] so occasionally it probably generates a fairly large delta [03:04] spiv: if we could fix the "re-read full inventories to generate the root texts" and "progress indication" I'd be happy to remove IDS :) [03:05] jam: I know I'd be happy to remove it :) [03:05] the progress thing is actually a pretty big deal [03:06] try upgrading something like even bzr using -DIDS:never [03:06] and it really just sits there for a long time [03:07] Yeah, I bet. [03:15] jam: we have to fix it for normal push pull too [03:15] I don't think it has to be a blocker; if we're going to fix it for 2.0. [03:15] spiv: at least for remote you get transport activity [03:16] removing IDS completely, it is a blocker for me [03:42] poolie: http://advogato.org/person/robertc/diary/107.html [04:02] poolie: iter-changes may become blocked soon; I'm waiting on Aaron's reply [04:04] lifeless: what 2 branches are they again? I'd like to run usertest on OOo on your work to date [04:05] hi all [04:11] iter-changes-partial-parents + commit-specific-files [04:30] I really wish review showed commit messages against lines [04:30] a mixed diff + annotate [04:30] * lifeless buginates [04:30] lifeless: what do you mean? [04:31] although on re-reading, I think I know [04:32] might be kinda hard in that we store the reviews as text... [04:32] diffs that is [04:55] hi, im getting an error- "bzr: ERROR: [Errno 13] Permission denied: '/home/alex/tangerine/tangerine-0.3.1/TangerinePrefPane/Makefile.in'" [04:56] that suggests that permission was denied :P [04:56] yes, im an idiot and failed to read to full path until right now [04:57] i managed to skip the tangerine-0.3.1 part of the path that make distcheck created [05:12] igc: did you find the branches? [05:13] lifeless: yet to look - will after I've finished my current email [05:17] hehe [05:18] the info page for bzr-gtk looks bad in bzr-gtk ;-P [05:31] hardlinking working copy files is not currently supported in [05:31] ?!?! [05:31] what changed? [05:39] thumper: you're now warned about it [05:39] mwhudson: when? [05:39] oh [05:39] before it just wasn't doing it? [05:39] aiui [05:40] ah [05:40] so that is where my disk is going... [05:41] thumper: yeah fraid so [05:41] NEWS has a pointer to the bug about it [05:41] you can comment there [05:46] thumper: https://bugs.edge.launchpad.net/bugs/408193 [05:46] Launchpad bug 408193 in bzr "bzr branch or checkout --hardlink has no effect in 2a format" [Medium,In progress] [05:47] poolie, ta === sidnei is now known as sidnei-away [06:16] Hello! Pardon me for a newbie question, but I'm curious about the best process for something. I'm building a web application, and as such I've got a 'development' test server and what I hope will be a 'production' web server. For the production side, should I use a branch and tagging mechanism, or is there a better way to do this? [06:23] KhaZ: kind of a big question but typically you'd have two branches and merge from development to production at intervals [06:23] whenever you roll out [06:23] spiv, how's it going? [06:24] poolie: enjoying the punishment that the combination of pushing part of mysql-server into 2a and using lsprof gives to my cpu :) [06:25] robert said he was going to review it after lunch.. [06:26] I am reviewing it [06:28] jam has sent a partial review too, where he says that his performance concerns have been addressed, which is good news. [06:44] poolie: is https://bugs.edge.launchpad.net/bzr/+bug/402778 the bug you wanted me to look at? [06:44] Launchpad bug 402778 in bzr "NoSuchRevision error when branching with rocketfuel-get" [Critical,Triaged] [06:44] actually no, though it does look useful [06:45] i was thinking of bug 375013 [06:45] Launchpad bug 375013 in bzr "Commit doesn't honor stacking invariants" [Critical,Triaged] https://launchpad.net/bugs/375013 [06:45] your choice [06:45] Ah, right. [06:45] I'll take a look, see which one appears easier ;) [06:46] I bags 375013, if you like [06:48] lifeless: sure, that makes sense I think. [06:49] As you've probably got commit's guts paged in already. [06:49] OTOH you've been doing streams most recently [06:50] have a look at the comment I put in 375013 recently [06:50] you may prefer to do it [06:58] hi all [07:13] I'd prefer my wireless to stop mysterious sucking. [07:13] Good afternoon vila. [07:15] spiv: yeah, hopefully the afternoon will be better than the morning with my daughter waking up at 39.5C :-/ [07:15] spiv: to give you an update [07:16] spiv: lots of little stuff; some baroqueness sneaking in that I think you can just totally axe. [07:16] vila: hopefully she's feeling better soon [07:18] igc: oh, I don;'t worry that much for her, except for the fact that she was about to fly for an additional vacation week and *that* will certainly be cancelled :-/ [07:18] lifeless: cool, that sounds promising. [07:26] I'm up to bzrlib/smart/repository [07:27] 20 pages to go [07:28] I'll be finishing it tomorrow [07:33] hi ho vila :P [07:39] pygi: hi ! [07:41] Hi guys, is there a command to pull and update at the same time? [07:46] rif_: no, folk either pull, or update :) [07:47] rif_: are you doing 'pull URL; update' ? if so, you shouldn't need to [07:48] lifeless: i was making a script to update several repos and I was wandering if i just missed that command [07:48] rif_: pull updates the working tree [07:48] update is used to update a tree against its own branch [07:49] lifeless: oaaaa [07:50] poolie: EOD; biab; [07:50] k, bye [07:51] lifeless: pull updates the working tree indeed :) [07:54] lifeless: so update would be used if someone pushed stuff to my branch to update the tree? [07:54] lifeless: http://gorf.tangent.org/hudson/job/Drizzle-subunit/3/ [07:56] hello vila [07:56] hi poolie [07:57] how's stuff? [07:59] kerguelen has been plugged as a slave, I'm debugging it right now [08:13] rif_: yes [08:13] is there a pointer to "magic that I can drop in .bzr/branch/branch.conf to run an arbitrary shell command on commit" - if indeed possible? Or an alternate method for same? [08:13] mwhudson: SEX [08:14] vila: http://gorf.tangent.org/hudson/job/Drizzle-subunit/3/ <- its drizzle, but thats what a result page could look like using subunit [08:14] lifeless: sexy indeed [08:15] mwhudson: oops, wrong nick :P [08:15] mtaylor: SEX [08:19] spm, post-commit plugin? :) [08:19] hook plugin* [08:20] spm: no, it would be a huge security hole [08:20] spm: you can however write a post commit or post_push or post_tip_change plugin in python, and have it get its parameters from branch.conf [08:22] vila: and here - http://gorf.tangent.org/hudson/ [08:22] you can see if you mouseover drizzle-subunit you get the test details [08:24] grpmf, definitely sexier than buildbot :-/ [08:34] lifeless: in this specific case, we'd be relying on O/S security to control who can read, let alone commit. Is that still a hole when used that way? [08:35] yes [08:35] if root ran 'bzr commit' in a different tree on the system, it could do arbitrary code as root. [08:36] the _facility_ to run arbitrary code on commit is the hole [08:36] doing a plugin as I described isn't a hole [08:36] Ah. I see. the old PATH=. exploit in essence. right. [08:37] yes [08:37] I've also jsut realised there's probably one in ~/.bazaar/plugins [08:37] if you 'sudo bzr'... without HOME getting set to roots HOME [08:39] hmmm. wouldn't that be a weakness of su and not getting an env set? taking the crud with you? [08:39] yes [08:39] but the default is the default [08:39] sure [08:40] oki, ta, will pass that back to the questioner. much appreciated. [08:41] we'd be happy to help write a bzr plugin - its very easy to check a couple of branch configuration variables [08:41] so the idea is, to have a plugin that (say) sms's [08:41] youd set my_foo_number = ... [08:41] and my_foo_other = ... [08:41] in branch.conf [08:42] then the plugin checks branch.get_config().get_user_variable('my_foo_other') [roughly, details vary with contact with the enemy] [08:43] in this case it's to trigger an earlier run of a cron task, but yeah [08:44] mwhudson: http://src.opensolaris.org/source/diff/opengrok/trunk/src/org/opensolaris/opengrok/analysis/executables/ELFAnalyzer.java?r2=%252Fopengrok%252Ftrunk%252Fsrc%252Forg%252Fopensolaris%252Fopengrok%252Fanalysis%252Fexecutables%252FELFAnalyzer.java%40554%3Aa40bc05ca4e6&r1=%252Fopengrok%252Ftrunk%252Fsrc%252Forg%252Fopensolaris%252Fopengrok%252Fanalysis%252Fexecutables%252FELFAnalyzer.java%40509%3Ae65b56e2a9f0 [08:44] mwhudson: pretty [08:57] Nice URL there [08:58] Love those percent signs [08:59] that part is scary yes [09:00] heh, the nice thing is they're escaped percent signs, serving as escape signs [09:01] Yeah, it's almost as much fun as 7 \'s in a row in the middle of a string :p [09:02] seven \' or seven \ ? [09:02] Yes. [09:02] tought so [09:03] Git has a revert command which has a -n option: [09:03] --no-commit don't automatically commit [09:04] what in god's name is that! [09:04] The way I see it, the problem is that you're looking too closely at git... [09:04] But git revert xyz == "bzr merge -rxyz..parent:xyz . ; commit" [09:06] (presumably, it's well-named, since it's pretty much synonymous with --no-commit on git merge) [09:12] Anybody home? === vila is now known as vila-doctor [09:17] anyone know of a how-to on setting up bazaar on a godaddy server with SSH access? I know it can be done [09:25] * igc dinner === Noldorin_ is now known as Noldorin [09:55] vila-doctor: when you get back - another hudson feature - graphs - http://gorf.tangent.org/hudson/job/Drizzle-subunit/ === Noldorin_ is now known as Noldorin [10:11] is anyone aware of a way to see the diff between the working tree and a pending update on a checkout? bzr status gives me 'working tree out of date, run bzr update', but i'd like to see what I'm applying.. any ideas? [10:12] bzr diff -r $(bzr revno)..-1 [10:12] ? === vila-doctor is now known as vila [10:15] lifeless: I'd be interested by their time-travel machine, the results in your last URL are for 2009081[23] 8-D [10:16] ;) [10:17] Kinnison: thanx, this shows the diff between the last and pre-last revs right? [10:19] sender: I'd expect "current" and "head" [10:19] sender: I'm guessing -- I don't tend to use checkouts like that [10:20] gnight [10:20] Kinnison: I think you are right: head has revno 132, bzr diff -r 131..-1 gives the diff for rev 132... very useful [10:21] Kinnison: I only use it for live instances of sites, as a lightweight checkout of a repo on the same server [10:22] Kinnison: only thing is, a working tree could be several revs 'out of date'.. [10:28] Kinnison: same issue: https://bugs.launchpad.net/bzr/+bug/318604 [10:28] Launchpad bug 318604 in bzr ""bzr diff" after "bzr push" does not display changes" [Undecided,New] [10:41] Hmm === Noldorin_ is now known as Noldorin [10:53] Kinnison: 'revno' gives the revision number of the _branch_, not the working tree. [10:53] Kinnison: You wanr 'revno --tree' for that. [10:56] sidnei-away: ping [10:59] fullermd: aah [10:59] * Kinnison only uses checkouts rarely :-) === sidnei-away is now known as sidnei [11:39] * sidnei yawns [11:39] wassup vila [11:40] sidnei: I change the AutoSSH to use '-M 0' instead of '-M 20000', it took me a bit of time to realize how to properly install the service... [11:40] aka using -u Admin [11:40] :-/ [11:41] anyway, that should address the connect/disconnect messages that kind of ruin the prettiness of the new .css :) [11:42] sidnei: also, I fixed one DNS IP address and add two more but still no cigar, I'll check with jam later [11:42] hi! I've just gotten a crash while doing 'bzr rebase': "bzrlib.erros.NoSuchRevision" [11:46] vila: ok, thanks for letting me know. do you need any more help? [11:47] is anyone able to help with my rebase crash? [11:47] sidnei: can you explain why 'nslookup host' can resolve a host when 'ping host' can't ? [11:49] sidnei: otherwise, there are a couple of things I'd like to address in the slave/builder definitions but I need to know how you create the slave and setup the environment, aka, more or less update the README for kerguelen [11:50] vila: i suspect it's too early in the morning for me to ponder about ping vs nslookup. :) [11:50] sidnei: ok, np, nearly lunch time here anyway, take your time :) [11:50] vila: as for the slave definitions, i just ran 'buildbot create-slave' afair, but the client is running a copy of buildbot trunk, so that might explain the difference [11:52] AAAARGH damn synergy virtualbox rdesktop windows incompatibility nightmare gimme my mouse back damn it [11:53] sidnei: to run start the slave via a service too ? [11:53] sidnei: do you start the slave via a service too ? [11:53] vila: yes, as explained here http://buildbot.net/trac/wiki/RunningBuildbotOnWindows [11:54] sidnei: I can' click on that anymore :-( I just can type here, will reboot shortly, if you can send me the relevant commands by mail, I'll update the README === Noldorin_ is now known as Noldorin [12:15] sidnei: back, [12:16] ok, reading the URL you mentioned, it appears that the slave is started with buildbot_service.py start "c:/" "d:/" [12:17] sidnei: What I did for all the others was to define the environment in the Makefile and start them with 'make start' so that env changes doesn't require restarting the service, but restarting the service takes the changes into account :) [12:19] vila: ok [12:20] sidnei: I don't know how much work it involves to get there but if you can start by putting the slave definition (.tac and info dir) under bzr that would be a good start :) [12:21] vila: yeah, i will do that today [12:21] ok ta [12:23] the idea above is to end up with a service that just do: 'make -C start' and nothing more [12:23] when I bzr log, is it possible to get the last log (-l1) before or at a given revision (-rX) ? [12:23] look at the other slave Makefiles [12:23] if I bzr log -rX and there is no change for that revision, I simply get nothing === vila is now known as vila-lunch [12:26] * fullermd can't parse "no change for that revision"... [12:27] The stuff log shows is properties of the revision, so they're there... [12:28] not if you do bzr log -rX foo.txt and foo.txt has not been changed in revision X [12:28] Oh. For that you'd want something like "bzr log -r..X -l1 foo.txt" [12:44] hey .bzrignore is typically added to the repo? [12:50] fxn: yes [12:57] It doesn't have to be; it works merely by being there. But most of the time, you add it, since it makes sense for anyone else or any other branches to have the same ignores. === mrevell is now known as mrevell-lunch [13:22] thanks! -r..X did the trick === Noldorin_ is now known as Noldorin [13:50] moin jam [13:50] morning jelmer [13:51] morning jam ! === vila-lunch is now known as vila [13:52] morning vila [13:58] hi! anybody known how to uninstall bazaar windows shell extensions(TortoiseBZR)? === mrevell-lunch is now known as mrevell [14:01] regsvr32 -u tbzr.dll [14:01] jam: LOL, parralel reviews :) [14:01] yeah, my mail got out before yours got in :) [14:01] jam: I think poolie has a plugin to change the default format to 2a and can experiment failures [14:02] jam: anyway, he'll tell us :) [14:02] well, it is easy enough to do it manually [14:02] given that it is a 1 line change [14:03] jam: but you should remember to undo it and do it in every branch, etc [14:10] are you supposed to be able to pull a subdirectory from a branch without pulling the full branch with bzr? [14:10] say "bzr branch http://foo.org/bzrbranch/some/sub/project"? [14:11] mobodo: no, that doesn't work [14:12] ahh, ok, so you'd have to make sub branches [14:12] mobodo: it may be supported at some point in the future, although there's nobody working on anything like that at the moment === Edwin is now known as Guest73773 === Guest73773 is now known as EdwinGrubbs === Noldorin_ is now known as Noldorin === Noldorin_ is now known as Noldorin [14:58] jam: I fixed some DNS entries on kerguelen this morning, but still no cigar [14:59] vila: do you have a traceback of the latest failure? [15:00] jam: the context is just trying nslookup and ping [15:01] i.e. 'nslookup host' now works (I think it didn't, can you confirm) but 'ping host' still doesn't [15:01] vila: I'm pretty sure 'nslookup' always works [15:01] as it goes and contacts a DNS server directly [15:01] whatever is messed up on Kerguelen doesn't effect nslookup [15:01] so 'ping' is the real test [15:01] ha ok [15:01] (I could be wrong, but I think that is what I found) [15:02] nslookup will use [the platform equivalent of] resolv.conf, but it does do all the query gen/send/recv/dissection internally, yes. [15:02] It doesn't use gethostbyname() or the like. [15:02] anyway, the first dns server in 'ipconfig /all' was wrong, I fixed it and added the 3rd and 4th [15:03] fullermd: we're talking windows here, do you know what the equivalent of resolv.conf is ? [15:03] Probably something in the registry. I'm pretty sure it's what ipconfig /all tells you. [15:06] fullermd: http://paste.ubuntu.com/248670/ , I find the IP routing Enabled: No suspicious as well as the default gateway being in a different subnet... [15:06] jam: by the way, why can't we use DHCP ? [15:06] DHCP for? [15:06] Eh. If it's not being used as a router, there's no real reason to enable routing. [15:07] jam: for IP address, default gateway, DNS [15:07] vila: last I checked the default gateway and subnet mask and default dns were a bit crazy [15:07] fullermd: even if talking to different subnets ? [15:07] when we asked the hosting provider to fix [15:07] they said they needed to wipe and re-install the os [15:08] :-( [15:08] if we are going to do that, we'll stop hosting with the [15:08] thm [15:08] vila: [I'm pretty sure] that means routing packets THROUGH the host; e.g., the host itself being a router. [15:08] them [15:08] Default gateway on a different subnet is unusual perhaps, but still valid (as long as routes exist to get the packets there of course) [15:09] But if it were wrong, you'd know it, 'cuz NOTHING would work, so that's not _the_ problem, whatever else it may be. === kiko-afk is now known as kiko [15:10] fullermd: Yeah, I thought a bit like that, but my past experiences with network is also that when I don't understand something it's often one or several bugs [15:10] s/often/often due/ s/bugs/wrong configurations/ [15:10] If nslookup works, that means all the network-related bits of DNS lookup is in good shape. [15:10] (well, nothing's _certain_, but that's about as good a sign as you can get) [15:10] so, backtrack, [15:11] if dns setup is right, what should be changed for ping to use it [15:11] So the likely place to something like ping to fail at getting the IP is in the resolver functions in libc. gethostbyname() etc. [15:11] I have no idea how that hooks together in windows, though. Presumably, it's the immoral equivalent of how it is on POSIX, but... [15:12] Does ping to bare IP work? [15:12] I thought their TCP/IP stack was based on BSD :-P [15:13] ping to hosts declared in the 'hosts' file works [15:13] Well, I'm based on amphibians, but I still can't hop worth a damn :p [15:13] and ping to bare IPs works too (just to answer the question) [15:14] Hm. Well, you've presumably got python on the box; you can try banging up a script to use python's gethostbyname() equivalent (which presumably hooks through libc), and see if you can get results back from that. [15:14] Or step down below that and throw together some C test code if you wanted. [15:16] fullermd: I was tempted to try that, but 1) asking first may have worked, 2) I strongly suspect that I will get the same results as ping (and I already encounter problems from python) [15:17] Oh. Well, in that case, I'd like to change my diagnosis to "It's messed up" :) [15:17] fullermd: gethostbyname fails [15:17] ping or direct access to IP addresses works [15:17] since we can set the IP in hosts and connect [15:17] (I tried that in the past :) [15:17] You tested down to the function level? [15:18] fullermd: your help is much appreciated (no kidding, walking the obvious (or not obvious) helps me keep my sanity) [15:19] Well, I have no real business doing any debugging on Windows; I haven't really touched it since I wiped it 95 my system 13 years ago :) [15:19] But DNS and networking and all, that I know. So hopefully I can bring a little something to the party anyway. [15:19] ... 13 years ago?? WTF? It can't have been that long... [15:19] fullermd: kind of same background here except I'm not *that* old :-D [15:20] Well, I didn't think *I* was either! [15:27] fullermd: I tested "py -c "import socket; socket.gethostbyname()" [15:27] * Tak only 11 years ago, youngun [15:29] Well, so that suggests it's b0rked at the python layer, and while it doesn't prove the problem is down in libc, the ping issue is symptomatically consistent with that, so it's probably a fairly safe guess. [15:30] (actually, it may be a separate libresolv and not actually in libc, but same difference) [15:30]