[00:26:45] ** _Maniac_ has joined us
[00:52:28] ** _Maniac_ has joined us
[02:07:17] ** gpciceri has joined us
[02:20:20] ** _Maniac_ has joined us
[02:47:45] ** _Maniac_ has joined us
[04:32:26] jack-e|away is now known as jack-e
[04:32:32] <jack-e> morning
[04:33:08] <vlado> good morning
[06:02:58] ** gpciceri has joined us
[07:46:45] ** vlado_ has joined us
[09:14:28] ** vlado_ has joined us
[10:13:33] _Maniac_ is now known as Maniac
[10:45:30] ** rdmurray has joined us
[11:39:05] ** _Maniac_ has joined us
[11:59:30] _Maniac_ is now known as Maniac
[12:09:13] ** gpciceri has joined us
[13:07:55] <Maniac> sqlite does support \describe to get a list of tables....
[13:11:30] ** jack-e has joined us
[13:11:59] ** gpciceri has joined us
[14:20:44] ** _Maniac_ has joined us
[14:21:43] jack-e is now known as jack-e|away
[14:35:21] ** vlado has joined us
[14:48:53] * vlado is away, somewhere far beyond... (l!on)(p!off) [cRk/bx]
[15:10:35] <rdmurray> maniac: So it does.  I think what happened was I tried it when I was having a problem, and it returned nothing, so I thought it didn't work.  Later, having forgotten about having tried \describe, I figured out I'd made a typo in the database name.  So that \describe I tried gave the correct output: there were no tables in the database I was accessing :)
[15:30:15] <rdmurray> There, the print references in helloworld are all fixed.
[15:30:58] <_Maniac_> wow my client died
[15:31:21] <rdmurray> The Internet hasn't been a very stable place for you lately, has it? :)
[15:31:25] <_Maniac_> grr
[15:31:31] <_Maniac_> no, i keep getting disconnected
[15:31:48] <_Maniac_> i've been chasing my ISP.. changed routers, generally complained alot.
[15:32:02] <_Maniac_> to make a short story shorter. they have no idea what the problem is
[15:32:23] <_Maniac_> anywho, what i typed into my client that never made it to the channel:
[15:32:42] <_Maniac_> i checked the source on sqlite. postgresql does not have \describe yet
[15:32:56] _Maniac_ is now known as Maniac
[15:34:20] <Maniac> so helloworld is becoming "the evolution of a peak application STARTING with helloworld"
[15:35:10] <rdmurray> Something like that :)
[15:35:41] <rdmurray> We could subtitle it, "A Story of Feature Creep"
[15:36:41] <rdmurray> I just found out that two of the examples don't work yet, presumably because PJE hasn't checked in the code.
[15:37:11] <Maniac> oo
[15:38:06] <rdmurray> * = commands.NoSuchSubCommand, and acceptURLs = False.
[15:41:08] <rdmurray> Hmm.  Time to go shovel snow off my roof
[15:41:10] <Maniac> recent cvs checkings
[15:41:15] <Maniac> cvs up!
[15:41:19] <rdmurray> I just did
[15:41:32] <rdmurray> Just before I typed that.  And I did a python setup.py install, too
[15:41:33] <Maniac> i thought i saw commit messages that referenced those ones...
[15:41:40] <rdmurray> There should have been, certainly.
[15:41:52] <Maniac> where do you live (that you have snow?)
[15:41:59] <rdmurray> New Hampshire
[15:42:14] <Maniac> ah, we have very little here still... Winnipeg,Manitoba Canada
[15:42:47] <rdmurray> Yeah, this one came up from the south.  Massechusettes actually got more.
[15:43:00] <rdmurray> Boy, I can't spell that state to save my ilfe.
[15:43:03] <rdmurray> life even.
[15:43:18] <rdmurray> Anyway, back in a few.
[16:00:06] ** _jpl_ has joined us
[16:00:34] <Maniac> hi _jpl_
[16:01:28] <_jpl_> <generic-greeting/>
[16:02:41] <_jpl_> <generic-query-on-well-being/>
[16:06:01] ** _jpl_ has left us
[16:09:55] <rdmurray> adapt(_jpl_,ILiveChatter).hello()
[16:10:56] <rdmurray> I decided to only do the (small) roof that was at the most risk from the coming rain.
[16:16:16] <vlado> NameError: name '_jpl_' is not defined
[16:34:41] <Maniac> TypeError: Invalid keyword argument for advising classes: moduleProvides
[16:34:44] <Maniac> yuk
[16:37:47] <vlado> you must call protocols.advise() with moduleProvides in module not in class 
[16:39:06] <vlado> for classes use instancesProvide or classProvides
[16:47:41] <Maniac> aha!
[16:55:39] ** gpciceri has joined us
[16:59:01] vlado is now known as vlado|zzzzzzzzzz
[17:02:30] ** pje has joined us
[17:02:43] <pje> rdmurray, I checked the code in for that stuff yesterday.
[17:03:26] <pje> See e.g. http://cvs.eby-sarna.com/PEAK/src/peak/running/commands.py.diff?r1=1.44&r2=1.45
[17:03:57] <pje> It was in CVS before I saved the changes to the page.
[17:07:28] <rdmurray> Hmm.  So why didn't I pick up the update :(
[17:09:13] <rdmurray> Hrmpf.
[17:09:27] <rdmurray> I re-did the cvs up and setup.py install, and now it works.
[17:09:41] * rdmurray decides to turn up the radio on that one.
[17:10:30] * Maniac can't get his plugins to work...
[17:10:34] <pje> Dunno.
[17:10:45] <pje> The anon CVS is updated every 5 minutes from the "real" CVS.
[17:10:49] <Maniac> i did the code as we discussed yesterday
[17:11:06] <Maniac> where in the class would i call binding.Make(plugins)
[17:11:31] <pje> def plugins(self):
[17:11:33] <pje>     ....
[17:11:39] <pje> plugins = binding.Make(plugins)
[17:11:54] <pje> Think of it as being like classmethod or staticmethod or property.
[17:12:13] <pje> It's a descriptor, just like them.
[17:12:54] <pje> rdmurray, what did you mean by "turn up the radio"?
[17:13:42] <rdmurray> Oh, that's an expression one of my old coworkers used.  It means something like "I'm going to pretend I didn't notice that anomolous behavior, in the hopes that it will never happen again"
[17:14:01] <Maniac> what error do you get if it can't adapt()?
[17:15:53] <Maniac> NotImplemented?
[17:16:00] <pje> Maniac, yes.
[17:16:06] <pje> NotImplementedError, to be precise.
[17:16:14] <Maniac> like what we have works but when i have a module that does not conform to the interface i get a traceback
[17:16:34] <pje> Well, there's a couple different ways to deal with that.
[17:16:49] <pje> E.g. adapt(aModule,IWhatever,None), and if m is None: continue
[17:17:07] <pje> Or..  adapt(aModule,IWhatever,NullPluginProvider)
[17:17:22] ** Maniac has joined us
[17:17:22] <vlado|zzzzzzzzzz> is there a way to advise() instances ?
[17:17:36] <pje> vlado|zzzzzzzzzz: protocols.adviseObject()
[17:17:47] <rdmurray> pje: After reading your updates, I tried to implement acceptURLs = False.  At first I tried to put that in the peak.commands.shortcuts section of the ini file, and it wasn't until that didn't work that I noticed that your text says to put it on the class.  I'm not sure whether or not that means the text should emphasize that point :)
[17:17:56] <vlado|zzzzzzzzzz> ah, thx
[17:18:17] <pje> Maniac, did you get the two ways I mentioned?
[17:18:42] <pje> There's also a third which is, "if people send you stuff that doesn't implement the interface, that *is* an error."
[17:19:11] <pje> Better that they get an error message right away, then have their plugins silently not work.
[17:19:16] <Maniac> yay!
[17:19:18] <Maniac>    def plugins(self):
[17:19:18] <Maniac>         plugins = []
[17:19:18] <Maniac>         for m in self.modules:
[17:19:18] <Maniac>             try:
[17:19:21] <Maniac>                 plugins.extend([factory() for factory in adapt(m,IBotPluginProvider).pluginFactories])
[17:19:23] <Maniac>             except:
[17:19:26] <Maniac>                 self.log.debug("%s does not appear to be a valid plugin" % m)
[17:19:29] <Maniac>             plugins.sort(self.pluginsCompare)
[17:19:31] <Maniac>         return plugins
[17:19:35] <pje> rdmurray: Yeah, it should be clearer, I guess.
[17:19:39] <Maniac>     plugins = binding.Make(plugins)
[17:19:39] <pje> Feel free to edit.  :)
[17:19:41] <Maniac> less lines or not isn't it more elegant?
[17:20:11] <pje> bare excepts are evil.
[17:20:27] <pje> You can catch errors that you don't intend to, like KeyboardInterrupt and such.
[17:21:05] <pje> The only kind of code you want to trap errors in are framework-level things like reactors or other event loops, anyway.
[17:21:16] <pje> Most everything else you want to break right away so that you know there's a problem.
[17:21:25] <pje> IOW, if you don't look at the debug log, you're toast.
[17:21:36] <pje> Heck, if you don't have debug logging enabled, you're toast.
[17:21:44] <pje> At least use log.error or log.warning.  :)
[17:22:29] <rdmurray> pje: meaning you aren't editing at the moment?
[17:22:39] <pje> rdmurray, No...
[17:22:49] <Maniac> pje, i think so (ie. get certainly the second)
[17:22:50] <pje> I've been busy with tools.supervisor bugs and related issues all day.
[17:23:39] <pje> Rolled out the new process manager to production today and all kinds of stuff broke.
[17:23:52] <pje> (Despite the week-long beta test.)
[17:23:54] <Maniac> you mean if it doesn't implement the interface is *should* traceback and die?
[17:24:12] <pje> Maniac, that's what I'm saying.  How will you know it's broken if it doesn't break?
[17:24:32] <pje> PEAK follows that philosophy throughout.  The sooner you get an error, the better.
[17:24:34] <Maniac> well, i should change to except NotImplementedError ?
[17:24:58] <pje> Maniac, how do you know the NotImplementedError wasn't raised in factory()
[17:25:00] <pje> ?
[17:25:06] <Maniac> not fair !
[17:25:32] <pje> What's not fair?
[17:25:44] <Maniac> you are tricky
[17:26:06] * Maniac is trying to think of any good reason to have modules in the modules directory that do not conform to the interface....
[17:26:07] <pje> What do you mean?
[17:26:16] <Maniac> or plugins
[17:26:29] <pje> I don't know how you get the modules.
[17:26:35] <pje> I assumed they were passed in by somebody.
[17:26:44] <Maniac> no, just globbing a directory
[17:26:49] <Maniac> (currently)
[17:27:33] <pje> Ah.  Well, in that case you probably want to use the adapt(...,None) or adapt(...,NullPluginProvider) approach.
[17:27:45] <Maniac> enlighten me sensei
[17:27:50] <pje> The latter means you need an object called NullPluginProvider with an empty pluginFactories list.
[17:28:29] <pje> e.g. class NullPluginProvider: pluginFactories=()
[17:28:33] <rdmurray> And that way your code just ignores modules that don't provide the interface.
[17:28:41] <rdmurray> If that's the behavior you want :)
[17:28:44] <pje> Right, it treats them like they just provide no modules.
[17:29:18] <pje> Er, no plugins.
[17:29:28] <rdmurray> I wish I had a faster gui browser than Mozilla.
[17:31:15] <Maniac> adapt(...,NullPluginProvider)  <---- what are the ....'s ?
[17:32:07] <Maniac> adapt(component, protocol, [, default [, factory]])   so i'm just providing a default?
[17:32:27] <Maniac> why not None?
[17:32:47] <pje> Because None doesn't have a 'pluginFactories' attribute.
[17:33:14] <pje> So you have to rewrite that nice list comprehension, or else use getattr.
[17:34:01] <pje> If you use a null object (like NullPluginProvider), you can just access the attribute.
[17:34:30] <pje> The best choice is highly context dependent.  There are places in PEAK that don't use a default, places that use None, and places that use a null object.
[17:35:06] <pje> Null object is most favorable when you have multiple methods or attributes on the interface.
[17:35:07] <Maniac> ah
[17:35:50] * Maniac doen't use list comprehensions unless they are handed to him on a platter
[17:38:12] <rdmurray> They aren't that hard, really :)
[17:38:52] <Maniac> no, i've just recently discovered them :)
[17:40:23] <rdmurray> ah.
[17:42:07] <rdmurray> Hmm.  Either I should have written these helloworld commands as IRerunnable's, or I've found a place to introduce pyprotocols, by writing a reverse adapter (ICmdLineApp -> IRerunnable).
[17:42:12] <Maniac> works quite well now pje
[17:42:37] <Maniac> and is more elegant and PEAKish at the same time
[17:43:39] <Maniac> the idea is to write plugins that work for IRC 'bots' and jabber 'bots' 
[17:43:53] <Maniac> so adaptation lovin is good
[17:45:08] <pje> You know, if you just use .ini or ZConfig files to specify plugins, you won't need to glob directories for modules.
[17:46:17] <rdmurray> Well, some people count being able to drop a plugin into a directory to "install" and active it as a feature :)
[17:46:33] <rdmurray> (I have mixed feelings, myself)
[17:47:21] <Maniac> but i want to be able to drop plugins in and reload them dynamically without restarting the bot
[17:47:25] * Maniac pokes rdmurray
[17:48:38] <pje> Maniac: then modules are a really lousy way to do that.
[17:48:51] <Maniac> teach me 
[17:49:01] <pje> You'd be better off with scripts that you run execfile() on.
[17:49:07] <pje> If you use modules, reloading is a problem.
[17:49:22] <pje> With scripts, there's no way for one script to depend on another.
[17:49:29] <pje> But with modules, one module can import another.
[17:49:41] <pje> When you reload, how will you be sure you've reloaded all the right modules?
[17:49:53] <Maniac> you have to be carefull?
[17:50:14] <pje> Say I have module A.  Module A loads module B.
[17:50:24] <pje> I notice that module B has changed on disk, so I reload it.
[17:50:25] <Maniac> no, i understand the issue
[17:50:33] <pje> But A is not reloaded.
[17:50:38] <Maniac> A still has reference to B
[17:50:53] <pje> Yep.  It could e.g. subclass a class in B.
[17:50:55] <Maniac> the modules must not reference each other
[17:51:16] <pje> Right; if you use scripts instead of modules, there's no easy way for them to reference each other.
[17:51:29] <pje> Except execfile, of course.
[17:51:42] <Maniac> here's a sample plugin containing module:
[17:51:44] <Maniac> http://randomthoughts.vandorp.ca/repos/projects/r
[17:51:46] <Maniac> grr
[17:51:48] <pje> OTOH, you could simply reload everything.
[17:52:09] <Maniac> http://randomthoughts.vandorp.ca/repos/sandbox/jabberbot/modules/hello.py
[17:52:49] <Maniac> i dont' think reloading everything is so simple is it?
[17:55:13] <pje> Not necessarily, no.
[17:55:35] <pje> Hence, my suggestion to execfile() them as scripts, instead.
[17:56:03] <pje> You exec 'em in a dictionary, then pull pluginFactories out of the dictionary, and throw it (the dictionary) away.
[17:56:28] <pje> If you want to restart the plugins, just throw away the old ones and re-exec the files.
[17:56:35] <pje> Of course, that brings up a point...
[17:56:55] <pje> If you were planning to reload them individually, how were you going to tell what module each plugin came from?
[18:05:59] * pje is gonna head out
[18:06:17] <pje> About to start a PC upgrade.
[18:06:23] <pje> Laters, all.
[18:06:30] <rdmurray> bye
[18:06:33] ** pje has left us
[18:17:30] <Maniac> well i had'nt figured out how to reload them individually yet :)
[18:21:46] <Maniac> mmm execfile could work i guess
[18:22:40] <Maniac> but then i still dont' know how you would reload individul pluginss.....
[18:23:01] <Maniac> you could re-execfile on 'modules'
[18:23:18] <rdmurray> that's what pje suggested, I believe.
[18:23:39] <rdmurray> Oh, individual
[18:24:08] <rdmurray> Well, you'd have to expand your plugins interface to support returning a source pointer, and record that somewhere, I suppose.
[18:50:23] <Maniac> yeah
[19:40:21] <rdmurray> Hmm.  Looks like there's a bug in my helloworld sql :(
[19:58:16] <rdmurray> Anyone know if peak's security system is only for the web stuff?
[20:02:25] <rdmurray> Hmm.  This is a short chapter.
[20:34:59] <Maniac> no, you can use peak.security for anything you please...
[20:35:13] <Maniac> http://peak.telecommunity.com/DevCenter/ShortIntroPeakSecurity
[20:37:10] <rdmurray> That doesn't seem to be linked from the front page.
[20:37:25] <Maniac> no, i wrote it and it hasn't been 'blessed' so i haven't linked it
[20:37:39] <rdmurray> Ah, I didn't realize things were supposed to get blessed.
[20:37:55] <Maniac> no, it's also incomplete (lacking an explanation
[20:37:56] <Maniac> )
[20:37:58] <rdmurray> Nor did I know how to create a page without linking it :)
[20:38:12] <Maniac> rdmurray, and they dont' have to be blessed :)
[20:40:29] <rdmurray> Do you know anything about the logging system?
[20:41:17] <Maniac> not very much just how to use the basic logger
[20:42:07] <rdmurray> Well, what I'm wondering about is this 'logging.logger:' business.  I don't get that at all.  'logfile:' I can understand, as well as the LogStream stuff.
[20:44:23] <rdmurray> Maybe I need to read PEP watzit.
[20:55:48] <Maniac> lazy binding is cool
[20:56:07] <Maniac> my 'bot' actually loads quicker now and doesn't load plugins until they try to get accessed
[20:56:28] <rdmurray>  Heh.  That's way cool :)
[20:56:32] * Maniac cheers for lazy binding
[20:56:44] <Maniac> i can imagine the difference this would make on 'real' applications
[20:56:54] <rdmurray> Yeah.
[20:56:58] <Maniac> since i'm still just playing / learning :)
[21:15:12] <rdmurray> have the changes to binding (ie: the existence of binding.Obtain) occur after alpha2, or are they in alpha2?
[21:22:53] <Maniac> this i do not know, check changes.txt
[21:23:04] <rdmurray> I am :)
[21:23:05] <Maniac> i believe a3
[21:23:47] <rdmurray> Yeah, looks like.
[21:36:11] ** pje has joined us
[21:36:24] <pje> rdmurray, we don't have syslog or logtee support ATM.
[21:36:47] <pje> That's why they're listed in the TODO section of the logs module.  :)
[21:37:57] <pje> Btw, I'd like to start splitting the tutorial into separate pages.
[21:38:36] <pje> I'm figuring I'll start a new page IntroToPeak, and give it sub-pages LessonOne, LessonTwo, etc.
[21:51:01] <Maniac> wb pje
[21:59:02] <Maniac> one big long page is intimidating
[22:02:25] <pje> Yep.
[22:02:47] <pje> I hope he isn't editing that page right now, since I'm busy carving it up into lessons...
[22:16:01] <Maniac> rdmurray!!!!!
[22:23:15] <pje> Okay, splitting is done...
[22:23:21] <pje> The new page is IntroToPeak
[22:23:50] <pje> Now if only I had time to keep working on the actual editing...
[22:24:06] <pje> He's already two and a half lessons ahead of me.  :)
[22:24:13] <pje> Gotta run; see y'all later.
[22:24:19] ** pje has left us
[23:55:48] ** hazmat has joined us