AOLserver Chat Logs

2004/07/28

IRC [00:41] *** xaqie joined the chat.
IRC [00:42] *** xaqie parted the chat.
IRC [04:52] *** frankie joined the chat.
AIM [05:34] *** POLearyUK joined the chat.
IRC [06:46] <Dossy> hey pat!
AIM [06:46] <POLearyUK> hey dossy, am sitting here with my replacement mike
AIM [06:46] <POLearyUK> am going to hook him up with you guys and aolserver chat room and stuff today
AIM [06:47] <POLearyUK> how's things in white plains?
IRC [06:49] <Dossy> dunno, i'm down in VA today :)
AIM [06:49] <POLearyUK> haha, sucker
AIM [06:49] <POLearyUK> D'oh
IRC [06:50] <Dossy> hehe
IRC [08:34] <AndyPiskorski> I to append a Tcl list element to a string, in C code. Like with Tcl_AppendElement but I want to append to an Ns_Dstring, not to the Tcl interp result. What C function should I use for that?? Anybody know?
IRC [08:38] <Dossy> well ...
IRC [08:38] <Dossy> you're going to have to create a Tcl list obj, then get the string rep of it
IRC [08:38] <Dossy> then shovel that into your Tcl_DString.
IRC [08:39] <Dossy> Tcl_ListObjAppendElement(interp, listPtr, objPtr)
IRC [08:39] <AndyPiskorski> Ah, ok. So I guess the existence of Tcl_AppendElement is a special case.
IRC [08:39] <Dossy> you want that.
IRC [08:39] <Dossy> you basically create a new list obj (Tcl_NewListObj) then Tcl_ListObjAppendElement() onto it.
IRC [08:40] <Dossy> then, Tcl_GetString() to get the string rep, which you can then create a Tcl_DString with.
IRC [08:42] <Dossy> The question is, if you already have the string rep of a Tcl list in a DString, how do you create a Tcl list obj from that efficiently ...
IRC [08:42] <AndyPiskorski> Hm, is it true that I never need to expliclty free the Tcl_Obj I'm creating here with Tcl_NewListObj()?
IRC [08:43] <Dossy> OMG
IRC [08:43] <Dossy> I'm smoking crack. Use Tcl_DStringAppendElement.
IRC [08:43] <Dossy> DUHHH
IRC [08:43] <AndyPiskorski> Oh, I don't already have the string rep of the Tcl list. I'm creating it in a loop.
IRC [08:43] <Dossy> Andy: you do have to free it UNLESS you're handing it back to the Tcl interp and telling it that you created it and want the interp to free it when refcount=0
IRC [08:44] <Dossy> well
IRC [08:44] <Dossy> is there a reason you're even at the C level? :)
IRC [08:45] <AndyPiskorski> Oh cool, Tcl_DStringAppendElement is exactly what I want.
IRC [08:45] <AndyPiskorski> Yes, I'm at the C level because I'm implementing 'ns_addrbyhost -all'.
IRC [08:45] <AndyPiskorski> All that DNS cacheing stuff AOLserver has makes it a bit more involved.
IRC [08:47] <Dossy> oh
IRC [08:47] <Dossy> please Tcl_Obj'ify the interface and create a proper Tcl_Obj list object please :P
IRC [08:48] <Dossy> time to disconnect and head into the office
IRC [08:48] <Dossy> be back on in a bit
IRC [08:48] <AndyPiskorski> Can't. The cacheing only works on strings, not on Tcl_Objs.
IRC [08:48] <AndyPiskorski> Maybe it should all be gutted and replaced with Tcl Thread Extension stuff, which probably can cache Tcl_Objs just fine.
IRC [08:48] <AndyPiskorski> But would be a bit behind what I'm prepared to do here just now...
IRC [08:49] <Dossy> i don't mean cache Tcl_Obj's
IRC [08:49] <Dossy> I just mean have it return a proper Tcl_Obj
IRC [08:49] <Dossy> plus, we need to investigate the thread-safety of gethostname_r() etc.
IRC [08:49] <AndyPiskorski> /*
IRC [08:49] <AndyPiskorski> * TODO: Wrapping a critical section around the maybe-not-thread-safe
IRC [08:49] <AndyPiskorski> * gethostbyaddr() and gethostbyname() is easy and should work
IRC [08:49] <AndyPiskorski> * everywhere, but it isn't exactly optimal.
IRC [08:49] <AndyPiskorski> *
AIM [08:49] <OnlineHost> You have been removed from the room for scrolling.
IRC [08:49] <AndyPiskorski> * On many operating systems, these functions are already thread-safe.
IRC [08:49] <AndyPiskorski> * On those where they are not, thread-safe alternatives like
IRC [08:49] <AndyPiskorski> * gethostbyname_r often ARE available, and when in that case (such as
IRC [08:49] <AndyPiskorski> * with Solaris/SunOS 5.8), writing our own ns_gethostbyname function
IRC [08:49] <AndyPiskorski> * using gethostbyname_r and thread-local storage is pretty easy.
IRC [08:50] <AndyPiskorski> * But, doing the configure script hackery for that might be harder -
IRC [08:50] <AndyPiskorski> * at least, I don't know how to do it.
IRC [08:50] <AndyPiskorski> *
IRC [08:50] <AndyPiskorski> * Perhaps even better, the Tcl Standard Library includes DNS
IRC [08:50] <AndyPiskorski> * functions. Are they thread-safe? Can we just use them instead
IRC [08:50] <AndyPiskorski> * here?
IRC [08:50] <AndyPiskorski> *
IRC [08:50] <AndyPiskorski> * --atp@piskorski.com, 2004/07/28 07:15 EDT
IRC [08:50] <AndyPiskorski> */
IRC [08:50] <AndyPiskorski> I don't know how to do the configure script magic for that.
IRC [08:50] <Dossy> ah, that's how the bot gets ejected from the chat room. argh.
IRC [08:51] <Dossy> nono - even gethostbyname_r() on solaris is not 100% MT
IRC [08:51] <Dossy> that's the problem
IRC [08:51] <AndyPiskorski> It's not??? How do you know? Isn't that what the damn "_r" functions are FOR?
IRC [08:51] <Dossy> anyhow - i'll catch up with you in a bit
IRC [08:51] <AndyPiskorski> Ok.
IRC [08:52] <Dossy> yes, but solaris has botched their libnsl/libresolv crap since solaris 2.6 :)
IRC [08:52] <Dossy> search sunsolve for bugs dealing with various get*byname() stuff
IRC [08:52] <Dossy> the fact that even into solaris 2.8, the resolver was single-threaded via mutex (!) to make it "thread-safe"
IRC [08:52] <Dossy> that it uses the FILE* interface and thus is limited to 255 open fd's even if FD_SETSIZE = 1024.
IRC [08:52] <Dossy> it's so badly broken. badly.
IRC [08:53] <Dossy> anyhow - enough rant. brb
IRC [08:53] <Dossy> :)
IRC [08:53] <AndyPiskorski> Well, I do know I had lots of mt-safety problems with 3rd party libs calling gethostbyname(), but I used LD_PRELOAD to override gethostbyname() with my own TLS version which called gethostbyname_r(), and that seemed to work fine.
IRC [08:53] <AndyPiskorski> Perhaps I was just lucky.
IRC [08:53] <AndyPiskorski> (That was on Solaris.)
IRC [08:54] <AndyPiskorski> [Leaving computer to eat breakfast now.]
IRC [09:12] <AndyPiskorski> Whoo! It seems to work! :)
IRC [09:16] <AndyPiskorski> Btw, when I had those thread-safety problems on Solaris way back when, I used LD_PRELOAD on several functions, not just gethostbyname(). So it's possible that gethostbyname() wasn't the one causing the symptoms I saw.
IRC [09:43] <Dossy> Possibl.
IRC [09:43] <Dossy> er. possibly.
IRC [09:43] <Dossy> and it seems gethostbyname_r() only has issues under load, or something.
IRC [09:44] <Dossy> and not necessarily limited to within the nsd. if total system-wide dns resolving activity is high, gethostbyname_r() seems to fail sometimes.
IRC [09:45] <AndyPiskorski> Yeah, that makes sense.
IRC [09:46] <AndyPiskorski> And of course that's why *NO* "thread-safe" library should EVER use mutexes around otherwise non-thread-safe function calls. (Unless, maybe, they also export the mutexes so they can be locked by foreign code.)
IRC [09:47] <AndyPiskorski> I have seen libs that say, "No, we're not thread-safe, but ALL you have to do is wrap a mutex around all calls to our library, and everything will be fine!" Not.
IRC [09:48] <AndyPiskorski> (Because that lib was calling non-thread-safe OS functions, which are of course called elsewhere too.)
IRC [09:49] <Dossy> well, it's true. but they mean *ALL* calls :)
IRC [09:56] *** rcrit joined the chat.
IRC [10:00] <rcrit> Dossy, you here?
IRC [10:04] <Dossy> yea
IRC [10:04] <Dossy> jsut responded to your mail :P
AIM [10:07] <POLearyUK> that's it documenting is the nemisis of all man kind!
IRC [10:07] <Dossy> what's wrong
IRC [10:07] <Dossy> documentation = http://google.com/
AIM [10:08] <POLearyUK> google://patricks work for past 5 years
AIM [10:08] <POLearyUK> result= 0 pages found
IRC [10:08] <Dossy> heh
IRC [10:09] <Dossy> then, you haven't done any! google knows all, and google doesn't like!
IRC [10:09] <Dossy> s/like/lie/
AIM [10:09] <POLearyUK> haha s/$.*^/$&/g;
AIM [10:11] <POLearyUK> no there is documentation, but it's a case that nobody can find it, as our document source has changed several times, and what people can find is usually first itteration
IRC [10:21] <rcrit> Dossy, forwarded you another mail to chew on
IRC [10:21] <rcrit> Mark reports that Tcl uses gethostbyname which isn't thread safe.
AIM [10:22] *** AIM IRC Bridge joined the chat.
IRC [10:24] <AndyPiskorski> The Tcl Standard Library isn't thread-safe? Ick.
IRC [10:25] <rcrit> yep, looks bad.
IRC [10:43] <Dossy> well
IRC [10:44] <Dossy> at least on solaris, gethostbyname() seems to call gethostbyname_r() ...
IRC [10:44] <Dossy> but yeah, at least inside AOLserver itself, I want to change calls to gethostbyname_r() directly -- just to be paranoid.
IRC [10:45] <rcrit> Dossy, no need, there are monitor locks around the calls
IRC [10:45] <rcrit> and yes on Solaris gethostbyname() calls gethostbyname_r but it uses a shared space which isn't thread safe
IRC [10:45] <rcrit> and there are several implementations of gethostbyname_r, some of which appear to be pretty buggy.
IRC [10:52] <Dossy> ok
IRC [10:53] <rcrit> its funny, this bug was opened against Tcl in 2000
IRC [10:53] <rcrit> along with some other un-thread-safe functions
IRC [10:53] <rcrit> and it was closed because no patch was provided :-(
IRC [10:59] <AndyPiskorski> That's awfully stupid. Can you re-open the Tcl bug?
IRC [10:59] <rcrit> I dunno, I'll try.
IRC [11:00] <Dossy> and lets submit a patch :P
IRC [11:01] <Dossy> grr, what the heck is wrong with the stock market these days :(
IRC [11:01] <AndyPiskorski> No matter what, "no patch provided" is not a valid reason for closing a bug.
IRC [11:01] <rcrit> Dossy: bleh
IRC [11:01] <AndyPiskorski> IMNSHO.
IRC [11:01] <rcrit> Yes, and I'm not an admin on that project and can't re-open bugs
IRC [11:02] <Dossy> andy: that may be "could not reproduce" as well.
IRC [11:02] <rcrit> they closed it as invalid
IRC [11:02] <Dossy> andy: using the non-threaded Tcl alone, the [socket] use of gethostbyname() won't ever produce a bug.
IRC [11:02] <rcrit> http://sourceforge.net/tracker/?group_id=10894&atid=110894&func=detail&aid=217833
IRC [11:02] <Dossy> rcrit: submit new, reference old bug ID at least
IRC [11:02] <rcrit> yes, that's what Imma gonna do
IRC [11:03] <Dossy> heh
IRC [11:03] <Dossy> "This stuff seems to vary a lot. Eg. on AIX4.3.2 the docs say that
IRC [11:03] <Dossy> gethostbyname is thread-safe, by gethostbyaddr is not. On AIX4.1.5
IRC [11:03] <Dossy> which we also have, both are thread-unsafe.
IRC [11:03] <Dossy> "
IRC [11:04] <Dossy> aches.
IRC [11:04] <Dossy> AIX. ACHES!
IRC [11:04] * Dossy mumbles something about having spent too many years fiddling with AIX 3.2.5 and Informix.
IRC [11:32] *** frankie parted the chat.
IRC [12:17] *** jader joined the chat.
IRC [13:17] *** GizmoBeastLives joined the chat.
IRC [13:21] <GizmoBeastLives> Dossy: I'm afaid the AIM bot got kicked out of the chat room again. By the way, how did you implement it? The old AIM archiving bot I wrote used Perl's Net::AIM module. Instead of implementing a smart way of having it reenter the chat room when it was kicked out, I had it automatically accept any chat invitations to the AOLserver room, that might be an easy option for your bot too.
IRC [13:35] *** bartt joined the chat.
IRC [13:45] *** frodoroot parted the chat.
IRC [16:29] <jhavard> I saw a book called "AIX for Breakfast" one day.
IRC [16:39] *** rcrit parted the chat.
IRC [17:17] <Dossy> giz: it's implemented in pure tcl --
IRC [17:18] <Dossy> what i need to do is just make the bot detect when it's been disconnected, and automatically reconnect to the chatroom.
IRC [17:18] <Dossy> It's not hard code to write, it's just hard to test. :)
IRC [17:18] <GizmoBeastLives> cool, is it all your own code or did you find someplace?
IRC [17:18] <Dossy> Because I can't force the AIM server to disconnect me.
IRC [17:19] <Dossy> It's all my own code, but loosely based off Perl's Net::OSCAR and all of the OSCAR doc for AIM and ICQ that's available freely on the net.
IRC [17:19] <Dossy> No privileged AOL information used.
IRC [17:19] <jhavard> wow.
AIM [17:19] <GizmoBeastLives> yes, I'm impressed
IRC [17:19] <jhavard> An AOL developer developing software for aol using non-aol sources.
IRC [17:20] <Dossy> jhavard: yah, because I wanted the potential to open source it once it's been cleaned up. the code is embarrassing still.
IRC [17:20] <Dossy> if I used AOL priviledged knowledge, I wouldn't be able to release it open source if AOL deemed it ungood.
IRC [17:20] <jhavard> ah.
IRC [17:20] <Dossy> or rather, I'd be at AOL's mercy.
IRC [17:20] <Dossy> but the way I did it, it's all my own.
IRC [17:21] *** GizmoBeastLives parted the chat.
IRC [17:21] <jhavard> so you mean to tell me you actually own what you create that isn't work related?
IRC [17:22] <jhavard> You mean to tell me that aol is less restrictive than this small isp I work for?
IRC [17:22] <Dossy> jhavard: I guess so.
IRC [17:22] <Dossy> jhavard: you own it if you work on it on your own time, on your own computers,e tc.
IRC [17:23] <jhavard> I don't.
IRC [17:23] <Dossy> See.
IRC [17:23] <jhavard> anything "within the scope of the companies business" is the company's
IRC [17:24] <jhavard> ack, I used plural instead of posessive. the grammar police will execute me.
IRC [17:24] <Dossy> when I came on board at AOL, I attached a rider to the non-disclosure/non-compete agreement listing all prior work I'd done on AOLserver as owned by me, so there'd be less chance they could claim ownership to any of it.
IRC [17:24] <thecodemill|aol> I'm logging. I don't understand 'I attached a rider to the non-disclosure/non-compete agreement listing all prior work I'd done on AOLserver as owned by me, so there'd be less chance they could claim ownership to any of it.', Dossy. Try /msg thecodemill|aol help
IRC [17:24] <Dossy> ok, that bot is annoying :)
IRC [17:25] <Dossy> ping
IRC [17:25] <Dossy> heh
IRC [17:27] <jhavard> well, anyways, I'm sad and depressed like an emo kiddie right now.
IRC [17:27] <Dossy> hey guys, are ANY of you pixel jockeys? or at least can anyone *write* marcom?
IRC [17:28] <Dossy> why?
IRC [17:28] <jhavard> whatawhat?
IRC [17:28] <jhavard> pixel jockey? marcom?
AIM [17:28] <GizmoBeastLives> I don't know what you're talking about either, so I'm guessing my answer is "no" :-)
IRC [17:29] <Dossy> okay - annoying: www.mail-archive.com shows up on google for the aolserver list more than gmane.org -- grrr
IRC [17:29] <Dossy> a pixel jockey is another term for webmonkey, or less perjoratively, "a graphic designer"
IRC [17:30] <Dossy> marcom is the common abbreviation for
IRC [17:30] <Dossy> no, marcom is the common abbreviation for "marketing communications"
IRC [17:30] <jhavard> Zhixel could probably use some cash.
IRC [17:32] <Dossy> is he any good?
IRC [17:33] <jhavard> http://www.hivehaus.com/
IRC [17:34] <Dossy> wow, not bad.
IRC [17:34] <Dossy> much better than I'd expected. hmm.
IRC [17:35] <jhavard> and don't forget his Howard "Hardcore" Dean photoshop job
IRC [17:35] <jhavard> http://www.kaiju.net/dean/
IRC [17:47] <Dossy> dude, we should recruit zhixel and build some wicked app for aolserver. grr.
IRC [17:47] <Dossy> is he working these days? maybe i can hire him :
IRC [17:47] <Dossy> :P
IRC [17:47] <jhavard> he's unemployed and looking for work, judging from his livejournal.
IRC [17:47] <jhavard> Wishes to get a job that will be enough work to warrant moving.
IRC [17:47] <jhavard> http://www.livejournal.com/users/zhixel/
IRC [17:48] <Dossy> i should ping him
IRC [17:49] <Dossy> better, he should join the chat here :P
IRC [17:56] <jhavard> Hrm... I wonder if I should use nsldap and store articles in ldap.
IRC [17:56] <jhavard> I've done a blog in ldap before.
IRC [17:57] <jhavard> and it worked quite well.
IRC [18:04] <Dossy> eek.
IRC [18:10] <jhavard> Ldap is read-oriented
IRC [18:10] <jhavard> most web stuff is read oriented
IRC [18:10] <jhavard> and you get threaded comments for free.
IRC [18:12] <Dossy> threaded comments for free? how do you figure?
IRC [18:18] <Dossy> brb, meeting
IRC [20:21] *** jader parted the chat.
IRC [22:18] <Dossy> anyone awake?
AIM [22:18] <GizmoBeastLives> I am
IRC [22:18] <Dossy> :)
IRC [22:18] <Dossy> I just got Jim's changes.
IRC [22:19] <Dossy> Gotta merge 'em in, but they'll be in HEAD soon enough!
AIM [22:19] <GizmoBeastLives> Great!
IRC [22:19] <Dossy> :P :P :P
IRC [22:19] <Dossy> *dances*
AIM [22:19] <GizmoBeastLives> I'm looking forward to trying those out
IRC [22:19] <Dossy> should be interesting. should also put a nail in the coffin for needing artblaster anymore
IRC [22:20] <Dossy> Although I have a feeling AOL will still use ArtBlaster, but that's just inertia.
AIM [22:20] <GizmoBeastLives> I tried out your win32 build script by the way - I think it is quite nice and could be a good replacement for the project files without two much additional work
AIM [22:21] <GizmoBeastLives> The main issue I see is debug vs. release, which the script doesn't currently address at all,
AIM [22:21] <GizmoBeastLives> Unfortunately, this requires conditionalizing where files get built, compile and linker options, etc.
IRC [22:21] <Dossy> yeah.
IRC [22:21] <Dossy> which is easy
IRC [22:22] <Dossy> BUILD=release or BUILD=debug, and conditionalize a few things
IRC [22:22] <Dossy> I don't like the "build output in separate dirs"
AIM [22:22] <GizmoBeastLives> Yes, I started looking into it, you'll have to move the command line config checks to happen before the other settings happen, which isn't hard
IRC [22:22] <Dossy> before the other settings happen? why?
IRC [22:23] <Dossy> it just depends on the command line order.
AIM [22:23] <GizmoBeastLives> If you don't build output in separate dirs, your obj files will replace each other
IRC [22:23] <Dossy> right
IRC [22:23] <Dossy> and with dependency checking, nothing should get built unless you "build clean"
AIM [22:24] <GizmoBeastLives> your make::flag CFLAGS set happens before you process anything set on the command line for example
IRC [22:25] <Dossy> aha.
IRC [22:25] <Dossy> yeah - true
IRC [22:25] <Dossy> that's easily done. all that make:: stuff gets moved to a make::init that gets invoked after cmdline args are processed.
IRC [22:25] <Dossy> but yes, it needs some work before it's 100%, but it DOES work :)
AIM [22:25] <GizmoBeastLives> in my local copy I added an extra setting QUIET=1 though maybe VERBOSE=1 would be more appropriate, that did all of the logging of the cmds that you commented out
IRC [22:27] <Dossy> hmm
IRC [22:27] <Dossy> you need to cvs up
IRC [22:27] <Dossy> i added DEBUG=1 :)
IRC [22:27] <Dossy> and wrap everything in if {$::DEBUG} { ... }
IRC [22:27] <Dossy> but I guess I could have added a QUIET= too --
AIM [22:29] <GizmoBeastLives> I'm using CVS HEAD, when did you commit?
AIM [22:29] <GizmoBeastLives> I haven't seen mail about it from the list,
AIM [22:29] <GizmoBeastLives> you use DEBUG=1 to signify logging instead of a debug build?
IRC [22:30] <Dossy> ahh shit
IRC [22:30] <Dossy> i didn't check it in.
IRC [22:30] <Dossy> yeah, DEBUG=1 signifies noisy build
IRC [22:30] <Dossy> BUILD=debug or BUILD=release is what I'd use to control the build type
IRC [22:30] <Dossy> and I guess I should add a "help" target --
IRC [22:31] <Dossy> so one could do a "build.tcl help" and get all the docs :)
AIM [22:31] <GizmoBeastLives> that to build help files or get help on using the tool?
AIM [22:32] <GizmoBeastLives> the ability to do both would be good of course :-)
IRC [22:33] <Dossy> hmmm yah. ugh
IRC [22:33] <Dossy> :)
IRC [22:34] <Dossy> OK, checked in.
IRC [22:34] <Dossy> now, lets work on merging Jim's changes
AIM [22:35] <GizmoBeastLives> I'll try to take a look tonight
IRC [22:35] <Dossy> one questionL: i started changing the autoconf to produce a config.h -- i'm having second thoughts now
AIM [22:36] <GizmoBeastLives> another question I had was if there is a particular reason for requiring tcl to be "installed" rather than just working from where the files are compiled
IRC [22:38] <Dossy> because aolserver needs them installed properly?
IRC [22:38] <Dossy> well, not really, but it's safer that way
IRC [22:38] <Dossy> because if you build shared, you definitely need the tcl shlibs.
AIM [22:39] <GizmoBeastLives> yeah, that was another thing I noticed, why are you linking against libc instead of msvcrt?
AIM [22:42] <GizmoBeastLives> I think it is more likely that people will want to specify a directory for both AOLserver and Tcl to install to
AIM [22:42] <GizmoBeastLives> like C:\nsd\nsd4-dev or something
IRC [22:43] <Dossy> am I linking libc?
IRC [22:44] <Dossy> I include the tcl dll's in the aolserver nsd bin/ dir
IRC [22:44] <Dossy> to ensure it's there and loaded
AIM [22:44] <GizmoBeastLives> It appeared that way
IRC [22:44] <Dossy> rather than the system searching PATH and finding a different/unusable tcl
AIM [22:44] <GizmoBeastLives> yes, that's where it belongs
IRC [22:44] <Dossy> no, it should be using msvcrt -- /MT is specified
AIM [22:44] <GizmoBeastLives> /MT -> LIBCMT.LIB
AIM [22:45] <GizmoBeastLives> I think the project files use /MD and /MDd
IRC [22:50] <Dossy> mm/
IRC [22:50] <Dossy> nope
IRC [22:50] <Dossy> they don't
IRC [22:50] <Dossy> and my system wouldn't build when i was missing msvcrt.lib
IRC [22:51] <Dossy> hmm.
AIM [22:51] <GizmoBeastLives> um, what line and which .dsp file?
AIM [22:51] <GizmoBeastLives> # ADD CPP, not # ADD BASE CPP, right?
IRC [22:51] <Dossy> look at win32/nsd/nsd.dsp
IRC [22:51] <Dossy> right
IRC [22:51] <Dossy> like line 76
AIM [22:52] <GizmoBeastLives> Those are the defaults, see line 77
AIM [22:52] <GizmoBeastLives> /MDd
IRC [22:52] <Dossy> ahh.
IRC [22:52] <Dossy> ok, i can make those changes -- next week
IRC [22:53] <Dossy> or week after
AIM [22:53] <GizmoBeastLives> yes, or maybe I'll find the time to beat you to it, you never know :-)
IRC [22:53] <Dossy> yeah, that'd be cool
IRC [22:53] <Dossy> and i want to eliminate all those dll link warnings too
IRC [22:53] <Dossy> try building with DEBUG=2 - ugly
AIM [22:54] <GizmoBeastLives> yeah, I've started working on scripting the build nightly on my website, it fetches the sources now anyway
AIM [22:54] <GizmoBeastLives> http://www.jamierasmussen.com/projects/source/
AIM [22:55] <GizmoBeastLives> I still have to set it up to run on a schedule instead of manually, and merge in a bunch of patches / modules from other places, etc.
IRC [22:56] <Dossy> interesting.
AIM [22:57] <GizmoBeastLives> I haven't added it in yet, but I think sloc provides an interesting summary of line counts by language and directory
IRC [22:57] <Dossy> hmm
AIM [22:57] <GizmoBeastLives> I ran it a while back and the modules I had were well over 223,000 loc
AIM [22:58] <GizmoBeastLives> I think the estimates it draws about man-hours etc. are on the iffy side, but just the breakdown by language etc. is probably pretty accurate
IRC [22:59] <Dossy> oh?
IRC [22:59] <Dossy> cool
AIM [22:59] <GizmoBeastLives> anyway, just one toy I've been thinking of playing with :-)
AIM [22:59] <GizmoBeastLives> having automated builds is soooo nice
AIM [23:01] <GizmoBeastLives> If you're still looking for roadmap ideas, automated builds and automated testing are two items I'd like to see in the near future
AIM [23:02] <GizmoBeastLives> Oh, have you thought about importing nstclprof into CVS?
AIM [23:03] <GizmoBeastLives> Would really hate to see it get lost again, even though it has some rough edges
IRC [23:06] <Dossy> yes
IRC [23:06] <Dossy> to automated tests
IRC [23:06] <Dossy> i'm actually working on getting AOL to pay for some
IRC [23:06] <Dossy> w/r/t/ nstclprof - I don't know if it's fully baked enough ...
AIM [23:07] <GizmoBeastLives> I wouldn't suggest putting it on the files page, but I don't think it can hurt to put it in CVS for bleeding-edge type people.
AIM [23:08] <GizmoBeastLives> I'll volunteer to do some pre-cleanup with respect to naming of file etc.
AIM [23:09] <GizmoBeastLives> I think it deserves to be there as much as nsftp does anyway :-)
IRC [23:10] <Dossy> heh
IRC [23:10] <Dossy> unfortunately can't name it nsprofile though :)
AIM [23:11] <GizmoBeastLives> yeah, that is an issue, I think nstclprof wouldn't be so bad
AIM [23:11] <GizmoBeastLives> Is the AIM IRC Bridge mimicing me?
IRC [23:12] <Dossy> i guess?
IRC [23:12] <Dossy> yeah, wow - weird
IRC [23:13] <Dossy> ping
AIM [23:13] <GizmoBeastLives> Actually, I didn't have a "But" at the front of my msg
IRC [23:13] <Dossy> ahh.
IRC [23:13] <Dossy> forget yeah, that
IRC [23:13] <Dossy> that's why
IRC [23:13] <Dossy> :)
IRC [23:13] <Dossy> it's the infobot feature. heh
IRC [23:22] <Dossy> you still there?
AIM [23:23] *** DossyNJ joined the chat.
AIM [23:23] <DossyNJ> still there?
AIM [23:23] <GizmoBeastLives> yes
AIM [23:23] <GizmoBeastLives> probably going to dinner soon though
IRC [23:25] <Dossy> ok
IRC [23:25] <Dossy> cool
IRC [23:25] <Dossy> yea, i'm gonna go to bed soon
IRC [23:25] <Dossy> gotta be up at 0530 am tomorrow to hop on the plane :(
AIM [23:25] <GizmoBeastLives> I'm sorry. Have a good flight
AIM [23:25] <GizmoBeastLives> Talk to you soon
IRC [23:26] <Dossy> thanks - have a good night
AIM [23:26] <GizmoBeastLives> good night
AIM [23:26] *** GizmoBeastLives joined the chat.