2005/02/28
IRC [02:27] *** wtem joined the chat.
IRC [02:28] <wtem> anybody around?
IRC [02:28] <wtem> debugging an issue with nsopenssl on mac os x with aolserver40r10
IRC [02:29] <wtem> then it will be on to debuggin my nsperm/tdav setup!
IRC [02:29] <wtem> hmm
IRC [02:29] <wtem> think i'll check openacs
IRC [02:39] <cbowman> what issues are you seeing with nsopenssl?
IRC [02:50] <wtem> ah
IRC [02:50] <wtem> still there?
IRC [02:51] <wtem> "nsopenssl: generating 1024 4-bit temporary RSA Key..."
IRC [02:51] <wtem> "Fatal: received fatal signal 10 Abort trap"
IRC [05:05] *** frankie joined the chat.
IRC [05:10] *** wtem parted the chat.
IRC [05:38] *** frankie parted the chat.
IRC [05:43] *** wtem joined the chat.
IRC [06:20] *** wtem parted the chat.
IRC [06:26] *** wtem joined the chat.
IRC [06:39] *** holycow parted the chat.
IRC [07:00] *** mcginniwa joined the chat.
IRC [07:00] *** wtem parted the chat.
IRC [07:23] *** mcginniwa parted the chat.
IRC [08:09] *** cbowman parted the chat.
IRC [08:28] <Dossy> Morning.
IRC [10:36] *** cbowman joined the chat.
IRC [10:42] *** wtem joined the chat.
IRC [10:58] <wtem> anyone had nsopenssl work with virtual serving under aolserver?
IRC [11:04] <wtem> also hoping to get tdav working with virtual serving...
IRC [11:14] *** mcginniwa joined the chat.
IRC [11:28] <jcdldn> wtem: Bartt has been using pound to do the ssl and according to him it works well
IRC [11:32] *** wtem parted the chat.
IRC [11:48] <mcginniwa> yeah
IRC [11:48] <mcginniwa> woah
IRC [11:48] <mcginniwa> nick changed
IRC [11:48] <mcginniwa> jcdldn
IRC [11:49] <mcginniwa> just had the realization the realization that pound might be the way to go while i was out walking the dog
IRC [11:52] <mcginniwa> know of any configuration notes for it?
IRC [11:53] <mcginniwa> specifically for the ssl and such?
IRC [11:53] <mcginniwa> ah, looks like https is native in pound
IRC [12:00] <mcginniwa> http://jamesthornton.com/writing/openacs-pound.html
IRC [12:00] <mcginniwa> might do the trick
IRC [13:25] <mcginniwa> i've built pound and it seems like it will do the trick
IRC [13:25] <mcginniwa> although i haven't tested it yet and there were some warnings during make
IRC [13:25] <mcginniwa> anyway
IRC [13:25] <mcginniwa> now my question is this
IRC [13:26] <mcginniwa> i want to enable admin webdav access to / on a server and restrict to GET/POST (and I think HEAD) methods to everyone else
IRC [13:27] <mcginniwa> is there a way of doing this with nsperm that doesn't require login for normal users?
IRC [13:27] <Dossy> sure
IRC [13:27] <Dossy> well, what do you mean?
IRC [13:27] <mcginniwa> i'll grab my perms file
IRC [13:27] <mcginniwa> hold on
IRC [13:27] <Dossy> if "normal users" attempt a HTTP method other than GET/POST/HEAD, it should prompt them for auth, no?
IRC [13:29] <mcginniwa> right
IRC [13:29] <mcginniwa> but otherwise no prompt
IRC [13:30] <Dossy> Hm, I wonder if WebDAV can do the right thing here.
IRC [13:30] <Dossy> are you using tDAV?
IRC [13:30] <mcginniwa> yeah
IRC [13:30] <mcginniwa> that was the idea
IRC [13:30] <mcginniwa> thinking i might have to do this:
IRC [13:31] <mcginniwa> admin.myserver.com with tdav and login has same pageroot as www.myserver.com
IRC [13:32] <mcginniwa> in other words two server instances, one with tdav and nsperm restricting access
IRC [13:32] <mcginniwa> the other without
IRC [13:32] <mcginniwa> seems like bloat
IRC [13:33] <mcginniwa> how would you write the perms file if it was without tDAV?
IRC [13:33] <Dossy> is the site high-traffic enough that bloat makes a difference, really?
IRC [13:33] <Dossy> keep things simple, you're more likely to get them right that way :)
IRC [13:33] <mcginniwa> # denyuser inherit GET / ""
IRC [13:33] <mcginniwa> denyuser inherit PUT / ""
IRC [13:33] <mcginniwa> # denyuser inherit HEAD / ""
IRC [13:33] <mcginniwa> ...
IRC [13:33] <mcginniwa> allowuser inherit GET / nsadmin
IRC [13:33] <mcginniwa> allowuser inherit PUT / nsadmin
IRC [13:33] <mcginniwa> allowuser inherit HEAD / nsadmin
IRC [13:34] <mcginniwa> ...
IRC [13:34] <mcginniwa> that's what i had in my perms
IRC [13:34] <mcginniwa> couldn't find a good nsperm doc when i looked really
IRC [13:34] <mcginniwa> tDAV seemed to barf on the inherit stuff
IRC [13:35] <mcginniwa> i turned that off to isolate nsperm
IRC [13:35] <mcginniwa> and then was prompted on every request to login
IRC [13:36] <mcginniwa> no, the site is not likely to be high traffic enough to make a difference for one
IRC [13:36] <mcginniwa> site
IRC [13:36] <mcginniwa> but i'm going to be running about six sites under pound
IRC [13:36] <Dossy> hmm
IRC [13:37] <mcginniwa> and to have webdav interface for non-tech admins for the homepage, etc is appealing
IRC [13:37] <Dossy> nsperm should have allowuser rules for GET/HEAD/POST for user ""
IRC [13:37] <mcginniwa> and having twelve servers seems crazy
IRC [13:37] <Dossy> and denyuser rules for all other DAV methods
IRC [13:37] <mcginniwa> ah
IRC [13:37] <Dossy> then allowuser rules for the DAV methods with a username other than ""
IRC [13:37] <Dossy> I think that'll work.
IRC [13:37] <mcginniwa> cool
IRC [13:38] <Dossy> but then the DAV code also needs to be smart and say "only allow GET/HEAD/POST DAV methods for users who are authenticated"
IRC [13:38] <mcginniwa> i'll give that a try
IRC [13:38] <Dossy> because DAV uses GET/POST/HEAD too, doesn' ti?
IRC [13:38] <Dossy> er doesn't it?
IRC [13:38] <mcginniwa> yep
IRC [13:38] <mcginniwa> but it's mainly put that i want to restrict for webdav
IRC [13:39] <mcginniwa> and move, copy, etc
IRC [13:40] <mcginniwa> ok, i'll go bang my head against this for awhile and see what i come up with
IRC [13:44] <Dossy> it's definitely possible, though
IRC [13:46] <mcginniwa> yeah, sounds like was just missing the allow stuff before
IRC [13:46] <Dossy> heh
IRC [13:46] <Dossy> cool.
IRC [13:47] <mcginniwa> virtual hosting may have confused tdav as well
IRC [13:52] <cbowman> Dossy: I'm seeing nsopenssl leaking SSL objects, has anyone else reported similar issues?
IRC [14:08] <Dossy> cbowman: yes, what version of nsopenssl?
IRC [14:08] <Dossy> there was a long thread investigating nsopenssl leakages and I *think* in the very latest beta and/or CVS HEAD, all known issues that could be found were fixed
IRC [14:09] <cbowman> 3.0beta26
IRC [14:10] <Dossy> where are you seeing leakage?
IRC [14:11] <cbowman> I'm seeing a leak of the memory allocated with SSL_new.
IRC [14:12] <cbowman> if I turn on debugging and add a few more Ns_Log lines in nsopenssl, I see two calls to NsOpenSSLConnCreate but only one to NsOpenSSLConnDestroy.
IRC [14:15] <Dossy> oh interesting
IRC [14:15] <Dossy> HOW are you getting that to happen? I did the same thing in my local copy (sprinkle Ns_Log at "checkpoints") and have only seen one-to-one
IRC [14:16] <Dossy> I've even checked the Tcl API stuff (ns_opensslsock* stuff)
IRC [14:19] <cbowman> the webpage I request has 2 images, so the browser opens a secondary connection.
IRC [14:20] <cbowman> that's the only thing I can think of that makes it even a little bit interesting.
IRC [14:21] <mcginniwa> Dossy: Choice, that allowuser did the trick
IRC [14:21] <mcginniwa> well for access without login anyway
IRC [14:21] <mcginniwa> now to test with tdav...
IRC [14:23] <mcginniwa> hmm
IRC [14:24] <Dossy> cbowman: hmm.
IRC [14:24] <Dossy> cbowman: could this be a HTTP Keep-Alive over SSL issue/
IRC [14:25] <Dossy> cbowman: aolserver 4.0.10, right?
IRC [14:25] <mcginniwa> Dossy: [28/Feb/2005:11:24:45][29541.2684396012][-main-] Error: tcl: source /web/mars-h\q/tcl/tDAV.tcl failed: invalid option "inherit": should be -noinherit
IRC [14:25] <mcginniwa> NONE
IRC [14:25] <mcginniwa> invalid option "inherit": should be -noinherit
IRC [14:25] <mcginniwa> while executing
IRC [14:25] <mcginniwa> "ns_perm denyuser inherit GET $uri """
IRC [14:25] <mcginniwa> invoked from within
IRC [14:25] <mcginniwa> "if { ![string equal "" $tdav_shares] } {
IRC [14:25] <mcginniwa> for {set i 0} {$i < [ns_set size $tdav_shares]} {incr i} {
IRC [14:25] <mcginniwa> set tdav_share [ns_configsecti..."
IRC [14:25] <mcginniwa> (file "/web/mars-hq/tcl/tDAV.tcl" line 2315)
IRC [14:25] <mcginniwa> invoked from within
IRC [14:25] <mcginniwa> "source $file"
IRC [14:25] <Dossy> right, by default it's inherit.
IRC [14:25] <Dossy> you override with -noinherit
IRC [14:25] <Dossy> heh
IRC [14:25] <mcginniwa> ah
IRC [14:25] <mcginniwa> ok
IRC [14:25] *** bartt joined the chat.
IRC [14:25] <mcginniwa> cool
IRC [14:26] <mcginniwa> hmm
IRC [14:26] <cbowman> Dossy: yes. it appears to be a keep-alive issue.
IRC [14:27] <cbowman> I see the following happen (added a couple debugs to NsSockClose):
IRC [14:27] <cbowman> [28/Feb/2005:19:22:19][19043.5][-conn:server1::0] Debug: NsOpenSSLConnCreate(0xb39d40)
IRC [14:27] <cbowman> [28/Feb/2005:19:22:34][19043.5][-conn:server1::0] Debug: NsSockClose: Closing Socket
IRC [14:27] <cbowman> [28/Feb/2005:19:22:34][19043.5][-conn:server1::0] Debug: NsSockClose: Keepalive state. Socket not closed
IRC [14:27] <cbowman> [28/Feb/2005:19:22:40][19043.7][-conn:server1::1] Debug: NsSockClose: Closing Socket
IRC [14:27] <cbowman> [28/Feb/2005:19:22:40][19043.7][-conn:server1::1] Debug: NsSockClose: Keepalive state. Socket not closed
IRC [14:27] <cbowman> [28/Feb/2005:19:22:40][19043.7][-conn:server1::1] Debug: NsOpenSSLConnCreate(0x1419fd8)
IRC [14:27] <cbowman> [28/Feb/2005:19:22:41][19043.5][-conn:server1::0] Debug: NsSockClose: Closing Socket
IRC [14:27] <cbowman> [28/Feb/2005:19:22:41][19043.5][-conn:server1::0] Debug: NsSockClose: Keepalive state. Socket not closed
IRC [14:27] <cbowman> [28/Feb/2005:19:22:42][19043.7][-conn:server1::1] Debug: NsSockClose: Keepalive state. Socket not closed
IRC [14:27] <cbowman> [28/Feb/2005:19:22:42][19043.7][-conn:server1::1] Debug: NsSockClose: Closing Socket
IRC [14:27] <cbowman> [28/Feb/2005:19:22:42][19043.7][-conn:server1::1] Debug: NsSockClose: Issuing a DriverClose call
IRC [14:28] <cbowman> [28/Feb/2005:19:22:42][19043.7][-conn:server1::1] Debug: NsOpenSSLConnDestroy(0xb39d40)
IRC [14:28] <cbowman> [28/Feb/2005:19:22:42][19043.7][-conn:server1::1] Debug: NsOpenSSLConnDestroy: SSL_free'd
IRC [14:28] <cbowman> [28/Feb/2005:19:22:42][19043.7][-conn:server1::1] Debug: NsOpenSSLConnDestroy: SSL References: 1
IRC [14:28] <mcginniwa> Dossy: the perms file looks like the argument are positional, should i replace inherit with ""?
IRC [14:29] <mcginniwa> seems like the tDAV.tcl code could just pull it out
IRC [14:49] <Dossy> mcginniwa: not sure ...
IRC [14:50] <Dossy> cbowman: Ah, so I'm right? This is only a leak with keep-alive -- ?
IRC [14:50] <mcginniwa> was a bug in the the tDAV.tcl ns_perm calls
IRC [14:50] <mcginniwa> dropped inherit in there
IRC [14:50] <mcginniwa> not in perms file
IRC [14:50] <Dossy> mcginniwa: interesting - you should talk to daveb about that in #openacs since he was the major contributor to tDAV, IIRC
IRC [14:50] <mcginniwa> yeah
IRC [14:51] <mcginniwa> think the tDAV.tcl code may have gotten stale since not it's in oacs-dav
IRC [14:51] *** bart1 joined the chat.
IRC [14:52] <cbowman> Dossy: from what I can tell, yes.
IRC [14:52] <Dossy> cbowman: this is helpful. maybe I can create a reproducible case and fix it - thanks!
IRC [14:52] <Dossy> did you also nail the nsoracle leak issue?
IRC [14:52] *** bart1 parted the chat.
IRC [14:52] *** bartt parted the chat.
IRC [14:54] <cbowman> yes and they are checked into cvs at sourceforge by jcollins.
IRC [14:54] <Dossy> awesome.
IRC [14:55] <mcginniwa> Dossy: good news is that i got tDAV and nsperm to work the way i want as far as permissions
IRC [14:55] <mcginniwa> bad news is that tDAV doesn't seem to do proper redirect from / to index.*
IRC [14:56] <mcginniwa> if you call /index.html explicitly it works
IRC [14:56] <mcginniwa> investigating
IRC [15:23] <cbowman> line 754 of ssl.c in nsopenssl says that NsOpenSSLConnFlush always returns NS_OK, but that's not true...
IRC [15:29] <Dossy> yeah.
IRC [15:29] <Dossy> nsopenssl needs more TLC -(
IRC [15:29] <Dossy> :-(
IRC [15:29] <Dossy> I've been slowly cleaning it up - Scott Goodwin seems to be really busy so I've kinda taken over maintaining it.
IRC [15:30] <Dossy> were you able to identify that leak w/ Purify's output alone?
IRC [15:30] <Dossy> if so, I seriously need to learn Purify :)
IRC [15:31] <cbowman> purify showed the leaks.
IRC [15:33] <cbowman> it shows a number of leaks, but most are small (and probably just one time leaks) so I've been ignoring them.
IRC [15:41] <cbowman> after I finally tracked down that it was leaking the entire ssl object, I started debugging the procs that create and destroy the objects and you know what I know now.
IRC [15:48] <Dossy> that's awesome;.
IRC [15:48] <Dossy> was it obvious that it was the SSL object that was leaking?
IRC [15:48] <Dossy> Can you email me the output from Purify so I get an idea of what it looks like?
IRC [15:50] *** holycow joined the chat.
IRC [15:54] <cbowman> I'll email you the output.
IRC [15:56] <cbowman> it isn't always obvious what is a real leak.
IRC [15:59] <Dossy> but, if it helped you find the leak in nsopenssl and nsoracle, that's great
IRC [16:03] <Dossy> wow, someone's building AOLserver on HP-UX. Pity them ... :-)
IRC [16:13] *** bartt joined the chat.
IRC [16:16] <mcginniwa> dossy:
IRC [16:16] <Dossy> ?
IRC [16:16] <mcginniwa> a way to get the name of the server from within a tcl script?
IRC [16:16] <Dossy> [ns_info server]
IRC [16:16] <mcginniwa> cool
IRC [16:17] <mcginniwa> thanks
IRC [16:19] <mcginniwa> damn i'm close
IRC [16:19] <mcginniwa> fixed that / issue
IRC [16:19] <Dossy> heh
IRC [16:19] <Dossy> cool
IRC [16:19] <mcginniwa> but now it looks to me like it may have screwed php
IRC [16:19] <mcginniwa> !
IRC [16:19] <mcginniwa> doh!
IRC [16:20] *** siddfinch parted the chat.
IRC [16:20] <mcginniwa> probably just forgot something in the config
IRC [16:22] <Dossy> and this is all on macosx? heh
IRC [16:24] <mcginniwa> yep
IRC [16:24] <mcginniwa> a straight install of openacs and all deps is easy with darwinports...
IRC [16:25] <mcginniwa> of course i gotta make it hard by throwing some nonstandard stuff into the mix!
IRC [16:29] <mcginniwa> yeah, tdav screws evaluation of scripts when you do "get"
IRC [16:29] <mcginniwa> frick
IRC [16:29] <mcginniwa> makes sense actually
IRC [16:30] <mcginniwa> if you are using a protocal like ftp, and you wanted to grab a php file, you would want to grab it's source code
IRC [16:31] <mcginniwa> not the result of eval
IRC [16:31] <mcginniwa> damn
IRC [16:32] <mcginniwa> hmm
IRC [16:33] <mcginniwa> gah
IRC [16:33] <mcginniwa> symlinks would be circular
IRC [16:34] <mcginniwa> hmm
IRC [16:37] <mcginniwa> this is what i'm thinking
IRC [16:38] <mcginniwa> one webdav.mydomain.com server with webdav dirs under pageroot that are symlinks to other server's pageroots
IRC [16:39] <mcginniwa> thoughts anyone?
IRC [16:41] <Dossy> try it and see :)
IRC [16:42] <tekbasse> is this so that only one webdav account is needed to maintain content on multiple sites?
IRC [16:42] <mcginniwa> no
IRC [16:43] <mcginniwa> so that admins can have access to pageroot via webdav
IRC [16:43] <tekbasse> oh.. i see
IRC [16:43] <mcginniwa> and that the pages will render correctly in a web browser for normal users
IRC [16:45] <mcginniwa> Dossy: yeah, i think i will
IRC [16:46] <mcginniwa> man, simplifying things can be soooo complex!
IRC [16:47] <holycow> that would make a good tshirt :)
IRC [16:51] <mcginniwa> it's the motto for OO programmers
IRC [16:51] <mcginniwa> heh
IRC [16:52] <mcginniwa> or perhaps for the people that billout for the programmers' time
IRC [16:56] <Dossy> mcginniwa: I'd check out Apache mod_dav and see how they do it/if it's possible/etc.
IRC [16:56] <Dossy> DAV clients may include something in the GET request that says "hi, I'm a DAV request - please give me the file contents, and not evaluate it" which tDAV should also notice.
IRC [16:58] <mcginniwa> yeah
IRC [16:58] <mcginniwa> i used mod_dav before
IRC [16:59] <mcginniwa> hmm
IRC [17:00] <Dossy> if the DAV protocol has no allowance for a client to distinguish itself from a normal web browser as part of its HTTP request, then you have NO choice but to divert DAV to a different server and/or to a sub-urlspace.
IRC [17:00] <jcdldn> I think the point is a get is a get is a get.
IRC [17:01] <Dossy> jcdldn: Not entirely true.
IRC [17:02] <mcginniwa> the interesting things is it looks like mod_dav defaults to the opposite
IRC [17:02] <mcginniwa> check out "i'mg getting php output rather than source" here:
IRC [17:02] <mcginniwa> http://www.webdav.org/mod_dav/install.html#complex
IRC [17:03] <mcginniwa> jcdldn: dossy is right
IRC [17:03] <mcginniwa> a get from ftp is not the same as get from web browser
IRC [17:03] <Dossy> >> haha
IRC [17:03] <Dossy> mcginniwa: I just found that snippet too
IRC [17:03] <Dossy> >> One common request is to use mod_dav to manipulate dynamic files (PHP scripts, CGI scripts, etc). This is difficult because a GET request will always run the script, rather than downloading its contents. One way to avoid this is to map two different URLs to the content, one of which will run the script, and one of which will allow it to be downloaded and manipulated with DAV.
IRC [17:03] <mcginniwa> dav's http is bastardization of ftp and http
IRC [17:04] <Dossy> so Apache's suggestion is sub-urlspace.
IRC [17:04] <mcginniwa> yep
IRC [17:04] <mcginniwa> seems like my webdav-admin server is an alright solution
IRC [17:04] <Dossy> It is, of course, the simplest solution and guaranteed to work unambiguously across all (even poorly) implemented DAV clients.
IRC [17:05] <mcginniwa> the kind of funky thing is when the server's being admin
IRC [17:05] <mcginniwa> 'd have a public webdav section1
IRC [17:05] <mcginniwa> but i think that should still work
IRC [17:06] <mcginniwa> guess i'll have some lunch and chew on it
IRC [17:06] <cbowman> Dossy: should I try TOT cvs for AOLserver to try to reproduce the bug? I noticed there are a number of changes from 4.0.10.
IRC [17:07] <Dossy> TOT?
IRC [17:07] <Dossy> top-of-tree?
IRC [17:07] <cbowman> yes
IRC [17:07] <Dossy> you can try, but I don't think it'll make a difference
IRC [17:08] <Dossy> and what's currently in HEAD may or may not be stable :)
IRC [17:16] <cbowman> looks like it doesn't parse adp pages and the cleanup issue still existed.
IRC [17:18] <Dossy> hmm - huh?
IRC [17:18] <Dossy> doesn't parse ADP pages? wha?
IRC [17:20] <cbowman> it return the adp without doing variable substitution beforehand.
IRC [17:20] <cbowman> might be something wrong with by build or config that caused it though.
IRC [17:24] <Dossy> yeah, likely the config would be my guess.
IRC [17:47] *** cbowman parted the chat.
IRC [17:53] *** cbowman joined the chat.
IRC [17:59] *** siddfinch joined the chat.
IRC [18:05] *** rubick parted the chat.
IRC [18:28] *** cbowman parted the chat.
IRC [18:47] *** holy_cow joined the chat.
IRC [18:48] <Dossy> I'm really happy with the increased level of participation from the non-AOL members of the AOLserver community. I'm not sure what triggered the shift from "AOL is repressing us" to whatever state we're in now, but I'm glad it happened.
IRC [18:49] <jcdldn> I am convinced it's your sparkling personality.
IRC [18:50] <Dossy> Thanks for the sarcasm. ;-P
IRC [18:50] <Dossy> Seriously, I wonder what finally made the shift happen. I'm thinking it was Jeff Hobbs' comment...
IRC [18:50] <Dossy> As much of a self-proclaimed egomaniac as I'd like to be, people don't respect me nearly as much as they respect Jeff.
IRC [18:51] <Dossy> Frankly, I don't respect me nearly as much as I respect Jeff: he's one of my role models, for sure.
IRC [18:51] <jcdldn> I think him weighing in helped -- I also think it was people finally realizing they could see things getting done.
IRC [18:51] <Dossy> Speaking of getting things done, Alfred hasn't gotten back to me about the new design for the aolserver.com site. Hmm.
IRC [18:52] <Dossy> And siddfinch, where are we with the "new" server for aolserver.com?
IRC [18:52] <Dossy> jcdldn: Strangely, I feel like I've gotten very /little/ done in the ~9 months I've been project leader.
IRC [18:52] <Dossy> It's actually something that's gotten me into a funk or mild depression lately.
IRC [18:53] <jhavard> Well, I don't see that many commit messages. I know that doesn't mean "no work"
IRC [18:53] <jhavard> but from a boss's view "more documentation of work" looks like work.
IRC [18:54] <jhavard> I can do one minor, inconsequential thing at work.
IRC [18:54] <jhavard> Generate tons of email, often discussing it with myself.
IRC [18:54] <jhavard> I'm a god.
IRC [18:55] <jhavard> I do 10 things in a day that would normally take two days to accomplish, but I only write about two lines in my daily report about it, I'm lazy.
IRC [18:56] <jcdldn> well I think clearing the backlog of tickets (mostly) and getting the communication from AOL going helps.
IRC [18:56] <Dossy> jhavard: yeah, I guess I internally measure my self-worth by the amount of code I commit. :-(
IRC [18:56] *** holycow parted the chat.
IRC [18:57] <Dossy> yeah, I'm still cutting through tickets, etc. -- it's hard to motivate to keep looking at those things. Especially those that I think "gee, anyone in the AOLserver Community could take this and do it ..."
IRC [18:57] <Dossy> It makes me feel like I'm robbing people of an opportunity to participate or contribute, and I know how AOL already had the outward appearance of not inviting community contributions ...
IRC [18:58] <Dossy> I guess people just need to be reminded that this IS my full-time job, so you'd expect AOL to have a majority contribution because of that fact.
IRC [18:58] <Dossy> So, nobody else here blogs?
IRC [18:59] <jhavard> Trying to find more motivation for your rss aggregator?
IRC [18:59] <Dossy> With my ns_xmlrpc thing, I'm thinking of putting together a blog hosting package for AOLserver. Since I totally *SUCK* at web UI design, I was going to just make an XML-RPC interface and suggest folks use Ecto or w.bloggar (which speak XML-RPC) as their UI, and use AOLserver for hosting/delivering blogs, RSS feeds, etc.
IRC [18:59] <Dossy> jhavard: that too! :-)
IRC [19:01] <jhavard> How many bugs have you closed out in the past year?
IRC [19:01] <jhavard> I mean there's 53 open bugs now.
IRC [19:02] <jhavard> and some aren't even in aolserver proper.
IRC [19:04] <Dossy> you know, SF's bug tracker reports are SO lame
IRC [19:06] <jhavard> how did the presentation go?
IRC [19:07] <Dossy> jhavard: awesome! went really well - turnout was 18-20 people, I think
IRC [19:07] <Dossy> maybe closer to 12-15 actually
IRC [19:07] <Dossy> folks thought I'd rehearsed the preso - was smooth apparently
IRC [19:07] <Dossy> and, I hate when people just read slides out loud so I hardly looked at mine :)
IRC [19:07] <jhavard> How many went, "Whoa. aolserver is neat?"
IRC [19:07] <jhavard> s/?/!/
IRC [19:08] <Dossy> not many, but I kept bashing Perl and Java :)
IRC [19:08] <Dossy> I showed the ns_register_proc example and said, "What I've done here in AOLserver/Tcl in 6 lines would have taken at least 3x-5x that in Java and perhaps 2x-4x in Perl.
IRC [19:08] <jhavard> what was the example?
IRC [19:08] <Dossy> of course with the "well-formatted lines" caveat.
IRC [19:09] <Dossy> it was borrowed from the Jim Davidson Digital City preso
IRC [19:09] <Dossy> ns_register_proc GET /demo/time getTime
IRC [19:09] <Dossy> proc getTime {} {
IRC [19:09] <Dossy> set page "<html><body>"
IRC [19:09] <Dossy> append page "The time is [clock format [clock seconds]]."
IRC [19:09] <Dossy> append page "</body></html>"
IRC [19:09] <Dossy> ns_return 200 text/html $page
IRC [19:09] <Dossy> }
IRC [19:10] <Dossy> Something trivially simple like that.
IRC [19:10] <Dossy> Obviously it could have been written:
IRC [19:10] <jhavard> Eh, you're looking at twice that for java.
IRC [19:10] <Dossy> proc getTime {} {
IRC [19:10] <Dossy> ns_return 200 text/html "<html><body>The time is now [clock format [clock seconds]].</body></html>"
IRC [19:10] <Dossy> }
IRC [19:10] <Dossy> But, I hate line wrapping source code on PowerPoint slides. :-)
IRC [19:10] <jhavard> perl could do it in just as many lines, but the perl mentality says it should end up being 4291 lines.
IRC [19:11] <Dossy> And it demonstrates how simple Tcl is to build up strings dynamically.
IRC [19:11] <Dossy> use CGI; my $cgi = new CGI; ... blah
IRC [19:11] <jhavard> tcl is essentially a lisp.
IRC [19:11] <jhavard> which I really, really like.
IRC [19:11] <jhavard> The first code I ever wrote was autolisp for autocad.
IRC [19:12] <jhavard> although I didn't really understand it.
IRC [19:12] <jhavard> Of course, I was 6 at the time.
IRC [19:14] *** cbowman joined the chat.
IRC [19:16] <jhavard> http://i.a.cnn.net/cnn/2005/LAW/02/28/jackson.trial/top.jackson.leave.ap.jpg
IRC [19:18] *** bartt parted the chat.
IRC [19:18] *** Starets joined the chat.
IRC [19:18] <Dossy> hey!
IRC [19:18] <Starets> hey Dossy.
IRC [19:18] <Dossy> pjansson@? is that the same Pete J?
IRC [19:18] <Starets> That would be me.
IRC [19:19] <Dossy> Wow, thought you bailed on the whole AOLserver thing
IRC [19:19] <Starets> I saw an interview with you recently, and you did well.
IRC [19:20] <Starets> I dont think I was adding much to the project
IRC [19:21] <Dossy> Starets: :-( Sorry to hear you felt that way.
IRC [19:21] <Dossy> (About your contribution to the project.)
IRC [19:21] <Dossy> Thanks for the positive feedback about the interview, though!
IRC [19:21] *** bartt joined the chat.
IRC [19:21] <Starets> I'm glad it's still going strong.
IRC [19:22] <Starets> Are you working from NoVA, or did you stay in NJ?
IRC [19:23] <Starets> (It was NJ, wasn't it?)
IRC [19:26] <Starets> Take care, Dossy. Say "hi" to Jim and company for me.
IRC [19:26] *** Starets parted the chat.
IRC [19:28] <Dossy> Ack - didn't get to respond. Yes, I'm in NJ ... :-)
IRC [19:39] <Dossy> I <heart> curl.
IRC [19:58] *** holy_cow parted the chat.
IRC [20:48] <cbowman> Dossy: when NsSockClose is called and the socket is marked as keep-alive (so the call to the driver's DriverClose), how does NsSockClose get called again so that the driver's DriverClose is called?
IRC [20:49] <cbowman> cause I see SockRelease gets called from the DriverThread when a socket gets into a timedout state, but I don't see how the driver's DriverClose ever gets called.
IRC [20:50] <cbowman> where "driver's DriverClose" == (void) (*sockPtr->drvPtr->proc)(DriverClose, sock, NULL, 0)
IRC [21:39] <cbowman> I'm thinking that in SockRelease the Reason_ReadTimeout case should include a call to the driver's DriverClose if the sockets keep flag is true.
IRC [21:54] <Dossy> hmm
IRC [21:54] <Dossy> yeah, that whole thing in 4.0 is gnarly
IRC [21:56] *** bartt parted the chat.
IRC [21:58] <cbowman> I'll have to agree.
IRC [21:59] <Dossy> stuff in 4.1 is much cleaner
IRC [21:59] <Dossy> keep-alive in 4.0 was hacked in - not very happy with it
IRC [21:59] <Dossy> hmm - looks like 95 bugs were closed in SF since 6/1/2004