From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Message-ID: <444758A2.8080403@altlinux.com> Date: Thu, 20 Apr 2006 13:47:14 +0400 From: Anton Farygin Organization: ALT Linux Ltd. User-Agent: Thunderbird 1.5 (X11/20060328) MIME-Version: 1.0 To: ALT Devel discussion list Content-Type: multipart/mixed; boundary="------------010900020203080700080307" Subject: [devel] [Fwd: Re: Better integration with power management scripts] X-BeenThere: devel@lists.altlinux.org X-Mailman-Version: 2.1.7 Precedence: list Reply-To: ALT Devel discussion list List-Id: ALT Devel discussion list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 20 Apr 2006 09:43:53 -0000 Archived-At: List-Archive: List-Post: This is a multi-part message in MIME format. --------------010900020203080700080307 Content-Type: text/plain; charset=KOI8-R; format=flowed Content-Transfer-Encoding: 8bit Они нашли ещё одно применение HAL'у ;) Rgds, Rider --------------010900020203080700080307 Content-Type: message/rfc822; name="Re: Better integration with power management scripts" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="Re: Better integration with power management scripts" Path: news.gmane.org!not-for-mail From: Richard Hughes Newsgroups: gmane.comp.freedesktop.hal Subject: Re: Better integration with power management scripts Date: Wed, 19 Apr 2006 18:47:19 +0100 Approved: news@gmane.org Message-ID: <1145468840.6102.4.camel@localhost.localdomain> References: <1145356883.2666.38.camel@localhost.localdomain> <1145404093.10216.29.camel@daxter.boston.redhat.com> Reply-To: richard@hughsie.com NNTP-Posting-Host: main.gmane.org Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="=-VRQkX4i+JbK6gbeRQ2gn" X-Trace: sea.gmane.org 1145468764 20162 80.91.229.2 (19 Apr 2006 17:46:04 GMT) X-Complaints-To: usenet@sea.gmane.org NNTP-Posting-Date: Wed, 19 Apr 2006 17:46:04 +0000 (UTC) Cc: HAL Development List Original-X-From: hal-bounces@lists.freedesktop.org Wed Apr 19 19:46:01 2006 Return-path: Envelope-to: gcfh-hal@m.gmane.org Original-Received: from gabe.freedesktop.org ([131.252.208.82]) by ciao.gmane.org with esmtp (Exim 4.43) id 1FWGkU-00056L-Lw for gcfh-hal@m.gmane.org; Wed, 19 Apr 2006 19:45:56 +0200 Original-Received: from gabe.freedesktop.org (localhost [127.0.0.1]) by gabe.freedesktop.org (Postfix) with ESMTP id 290259E825; Wed, 19 Apr 2006 10:45:53 -0700 (PDT) X-Original-To: hal@lists.freedesktop.org Delivered-To: hal@lists.freedesktop.org Original-Received: from uproxy.gmail.com (uproxy.gmail.com [66.249.92.169]) by gabe.freedesktop.org (Postfix) with ESMTP id B64759E79D for ; Wed, 19 Apr 2006 10:45:50 -0700 (PDT) Original-Received: by uproxy.gmail.com with SMTP id m2so864753ugc for ; Wed, 19 Apr 2006 10:45:49 -0700 (PDT) DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:subject:from:reply-to:to:cc:in-reply-to:references:content-type:date:message-id:mime-version:x-mailer; b=TTKI7t3izpp5SNtZDG2jclpR/Eu1lOzqeCbgnK4GFqMcpSi9uT25+dn1bWANRajivSs6xZmfxCqwKqXeTpQ7CzY1vivJFlzdwpR2toD9LOvJaz/sw75UhvJxiqHkbQD7TyoNQTOXfeS7SnfE1X/8ecksFsPV84gEX/U/Z4uOoBE= Original-Received: by 10.66.234.9 with SMTP id g9mr1241641ugh; Wed, 19 Apr 2006 10:45:49 -0700 (PDT) Original-Received: from ?10.0.0.14? ( [86.138.101.9]) by mx.gmail.com with ESMTP id k2sm157314ugf.2006.04.19.10.45.42; Wed, 19 Apr 2006 10:45:43 -0700 (PDT) Original-To: David Zeuthen In-Reply-To: <1145404093.10216.29.camel@daxter.boston.redhat.com> X-Mailer: Evolution 2.6.1 (2.6.1-1.fc5.2) X-BeenThere: hal@lists.freedesktop.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: hal technical discussion List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Original-Sender: hal-bounces@lists.freedesktop.org Errors-To: hal-bounces@lists.freedesktop.org Xref: news.gmane.org gmane.comp.freedesktop.hal:4907 Archived-At: --=-VRQkX4i+JbK6gbeRQ2gn Content-Type: text/plain Content-Transfer-Encoding: 7bit On Tue, 2006-04-18 at 19:48 -0400, David Zeuthen wrote: > On Tue, 2006-04-18 at 11:41 +0100, Richard Hughes wrote: > > Further to David's ideas, here are some of my ideas (shamelessly taken > > from the ubuntu pmi and fedora pmscripts): > > > > Create a /usr/share/hal/scripts/pm/hooks directory that distros and > > vendors can dump files into that do things on resume and suspend, for > > example: > > > > * unloading certain modules and re-modprobing them if required > > My position is that if a module needs to be unloaded and loaded again > there is a bug in the module that needs to be fixed. We should never do > this but providing the infrastructure for OS vendors, hardware/driver > vendors and users is fine. After all, we all know it takes time to fix > this. After a *long* chat on IRC, some more concrete ideas: * Get methods for suspending and resuming the display integrated with HAL as we can use fdi files to pick and choose how to resume each type of adapter. Something like video.tool_method (string) = "radeontool" and then all we have to do is call VideoResume() on the adapter, which launches the hal-system-video-resume script which does the resume (like the lcd brightness one). * Make hal depend on project pmt[1] unconditionally. * Make project pmt integrate with hal-system-power-* scripts, and get rid of the pm-utils, pmi, and PowerSave stuff in the existing scripts. Please see the attatchment of the IRC log, and you can see some of the rational (and arguments). Project pmt: What is it? * Provides simple shell command line tools to suspend and hibernate computer that can be used to run vendor or distro supplied scripts on suspend and resume. Does not depend on HAL. What isn't it? * There are no permission checks, no use of C, and certainly no DBUS! Why do we need it? * As all the main distros are re-implimenting the same thing, over and over, to do something that's really quite trivial. Having the common locations for stuff means that software and hardware vendors can just install one file to do the clever stuff that pmi & acpi-support (Ubuntu) and pm-utils (Redhat) are doing. HAL will use these scripts to perform the Suspend() and Hibernate() methods and then we can remove all the legacy cruft where we check for per-distro tools. * Distros can easily add/remove functionality by installing/removing one file into the hooks directory, for example: - enabling and disabling standby LED's on laptop hardware - enabling suspend GUI's like suspend2 - re-enabling video (perhaps using HAL, see above) - starting and stopping services that can't cope with suspending - re-syncing the time with ntp - removing and modprobing modules when needed - setting grub to be the default target for a hibernate-resume - other wacky things that need doing on specific systems * Perhaps apply for freedesktop for project pmt, or use the existing pm-utils codebase available at: CVS :pserver:anonymous@cvs.fedora.redhat.com:/cvs/devel, module is "pm-utils" * The future: When every module and program can deal with a suspend-resume cycle, and all the LED's work, and stuff doesn't break, then this module is obsolete. Until utopia arrives, I think this is required. [1] pmt is a silly name. We need to decide on a better one, perhaps pmi or pm-utils even. Discuss: :-) Richard --=-VRQkX4i+JbK6gbeRQ2gn Content-Disposition: attachment; filename=#hal.log Content-Type: x-directory/normal; name=#hal.log Content-Transfer-Encoding: quoted-printable Apr 19 13:24:12 * Now talking on #hal Apr 19 13:24:12 * Topic for #hal is: development discussion about fd.o ha= l - not an end-user support channel | http://www.chucknorrisfacts.com/ Apr 19 13:24:12 * Topic for #hal set by davidz at Fri Feb 10 18:52:51 200= 6 Apr 19 13:24:12 * Notify: dannyK is online (clarke.freenode.net). Apr 19 13:40:53 hughsie mjg59, can you tell me what you think of these: h= ttp://www.hughsie.com/applications/Coppermine/thumbnails.php?album=3D9 Apr 19 13:42:43 mjg59 hughsie: I think we need to spend some time hammeri= ng out the semantics of the inhibit interface Apr 19 13:42:56 hughsie mjg59, go-on... Apr 19 13:43:19 mjg59 I'd also argue that the statistics stuff is a bit c= rack-addled for g-p-m and should really be in a separate application, but= , well :) Apr 19 13:43:53 mjg59 hughsie: So I don't think it should be possible for= an application to block suspend - merely to delay it Apr 19 13:43:59 hughsie well, we use the data for the different algorithm= s, so it;s trivial to share. Apr 19 13:44:10 hughsie but yes, i get you rpoint Apr 19 13:44:23 hughsie why not block completely? Apr 19 13:44:36 mjg59 Because if I close my laptop lid, I expect it to su= spend Apr 19 13:45:02 hughsie but what if nautilus is copying files? Apr 19 13:45:18 mjg59 Then Nautilus gets to stop copying files Apr 19 13:45:24 hughsie :-) Apr 19 13:45:54 hughsie you seen the python example? Apr 19 13:45:58 mjg59 Nope Apr 19 13:46:02 mjg59 I'll look at that later on Apr 19 13:46:21 mjg59 But more to the point, I think we need to make sure= that we have an interface that everyone is happy with Apr 19 13:46:26 mjg59 Which includes the KDE and embedded people Apr 19 13:46:43 hughsie http://pastebin.com/669094 Apr 19 13:46:49 hughsie yes, i agree with that Apr 19 13:47:03 hughsie i figured it's better to start to do, and get an = idea of how it could be used. Apr 19 13:47:13 hughsie i'm open to changing names, or the way things are= done. Apr 19 13:47:35 mjg59 It turns out that embedded people have a different = set of things they want, such as "Please make the CPU a bit faster" Apr 19 13:47:55 hughsie in what situation? Apr 19 13:48:24 mjg59 So in desktop/laptop situations, we tend to allow a= pplications to grab as much CPU as they want to Apr 19 13:48:41 mjg59 In the embedded world, you have different constrain= ts (such as heat dissipation) Apr 19 13:48:59 hughsie sure, butthis is differnt to the inhibit stuff ri= ght? Apr 19 13:49:03 mjg59 Right Apr 19 13:49:13 mjg59 But it's part of the application interface to power= management Apr 19 13:49:22 hughsie ohh i see Apr 19 13:50:04 mjg59 Doing all of this via dbus makes sense, but it's pr= eferable to have a consistent API rather than a namespace with a bunch of= things that have been added over time Apr 19 13:50:15 hughsie i think the spec should start small tho, just to = avoid the 125% problem... Apr 19 13:50:25 mjg59 Right Apr 19 13:50:27 hughsie this OSI vs. Internet. Apr 19 13:50:33 hughsie *think, sorry Apr 19 13:50:44 mjg59 Pat Mochel is supposed to be setting up a forum whe= re we can discuss this with embedded people Apr 19 13:50:48 mjg59 And get a better idea of the problem space Apr 19 13:51:01 hughsie what list? Apr 19 13:51:09 mjg59 Doesn't exist yet Apr 19 13:51:12 hughsie ahh, i see Apr 19 13:51:20 mjg59 We only decided this on Friday :) Apr 19 13:51:25 hughsie cc me pls when he sets it up. Apr 19 13:51:47 mjg59 Sure Apr 19 13:52:02 mjg59 Also, Maemo is supposed to have an API for this sor= t of thing right now Apr 19 13:52:43 hughsie ohh, right. got a link or shall i google? Apr 19 13:53:22 mjg59 I haven't been able to find docs Apr 19 13:53:53 * dannyK is back. Apr 19 13:54:14 hughsie googling now Apr 19 14:01:20 mjg59 Ah Apr 19 14:01:24 mjg59 http://www.maemo.org/platform/docs/api/libosso/html= /index.html Apr 19 14:01:27 mjg59 Modules/device state Apr 19 14:01:39 mjg59 Looks fairly straightforward Apr 19 14:17:16 * kay (n=3Dkay@c147113.adsl.hansenet.de) has joined #hal Apr 19 14:22:22 hughsie mjg59, cheers for the link, i'll read it properly= later Apr 19 15:09:27 * ervin has quit (Read error: 104 (Connection reset by pe= er)) Apr 19 15:30:43 pjones hughsie: cvs is at :pserver:anonymous@cvs.fedora.r= edhat.com:/cvs/devel , module is "pm-utils". Apr 19 15:34:43 hughsie pjones, nice one, thanks. what did davidz say? Apr 19 15:35:01 pjones I haven't seen him since we spoke Apr 19 15:35:54 hughsie i think he still wants this in hal, if I readhis = email right. Apr 19 15:37:18 hughsie i have to revise, okay to talk later? Apr 19 15:37:45 * You are now known as hughsie_away Apr 19 15:51:13 * ervin (n=3Dipsquad@kde/ervin) has joined #hal Apr 19 16:14:50 * davidz (n=3Ddavidz@dhcp-15-40.media.mit.edu) has joined= #hal Apr 19 16:14:50 * Notify: davidz is online (clarke.freenode.net). Apr 19 16:15:03 davidz morning kay Apr 19 16:15:19 kay hey david! Apr 19 16:19:34 davidz kay: leaving tomorrow? Apr 19 16:20:18 kay yeah, 8am, it's in the middle of the night :) Apr 19 16:20:51 * davidz "=E2=99=AC=E2=99=AC leaving on a jet plane" :-) Apr 19 16:20:57 davidz how rude!! Apr 19 16:20:59 davidz 8am!! Apr 19 16:21:02 davidz don't miss your flight Apr 19 16:21:15 * davidz was delayed one day when I had an 8am flight fro= m Copenhagen to Bangalore Apr 19 16:21:24 davidz cuz I missed my connection in LHR Apr 19 16:23:50 kay davidz: yeah, should be fine this time, i'll get a ni= ce wakeup service :) Apr 19 16:24:00 davidz ooo Apr 19 16:24:02 davidz that's nice Apr 19 16:24:41 hughsie_away davidz, http://www.hughsie.com/applications/= Coppermine/displayimage.php?album=3D9&pos=3D2 Apr 19 16:24:51 davidz hi hughsie_away=20 Apr 19 16:24:53 * davidz looks Apr 19 16:24:57 hughsie_away a bit better than the old one.. :-) Apr 19 16:25:09 hughsie_away davidz, you had a chance to talk to pjones? Apr 19 16:25:13 davidz rad!! Apr 19 16:25:15 davidz no Apr 19 16:25:21 davidz I haven't talked to pjones Apr 19 16:25:30 * pjones is reading your mail from last night right now Apr 19 16:25:31 davidz and I wont be in the Red Hat office until next wee= k Apr 19 16:25:41 * davidz at OLPC headquarters Apr 19 16:25:46 hughsie_away lets let pjones catch up. Apr 19 16:25:47 davidz walking-distance from my home Apr 19 16:25:49 davidz yea Apr 19 16:26:03 hughsie_away richard at revision headquarters. Apr 19 16:26:32 hughsie_away revision blows goats. Apr 19 16:26:51 davidz revision? Apr 19 16:26:54 davidz is that a company name Apr 19 16:27:02 * davidz thinks of BBC's "The Office" :-) Apr 19 16:27:09 pjones davidz: ok, so you've got a big handwavy vagueness= here. Apr 19 16:27:17 pjones what, exactly, do you want to put in the fdi files= ? Apr 19 16:27:32 davidz information on how to resume video for instance? Apr 19 16:27:38 davidz look at the Ubuntu infrastructure Apr 19 16:27:42 davidz they have all that Apr 19 16:27:44 pjones url? Apr 19 16:27:45 davidz in a weird file format Apr 19 16:27:47 davidz uhh Apr 19 16:27:52 davidz hughsie_away, do you know the URL? Apr 19 16:28:05 * davidz looked at it some time ago Apr 19 16:28:06 hughsie_away weird, agreed, and no, but it'll be in my hi= story Apr 19 16:28:09 davidz like 3-4 weeks Apr 19 16:28:52 pjones what's the package name that I should look for? Apr 19 16:29:01 hughsie_away pmi Apr 19 16:29:02 davidz hold on Apr 19 16:29:04 davidz let me check Apr 19 16:30:22 davidz nah, I don't have it on my machine anymore Apr 19 16:32:38 * davidz needs to go to a meeting - bbiab Apr 19 16:32:50 hughsie_away http://archive.ubuntu.com/ubuntu/pool/main/p= /powermanagement-interface/powermanagement-interface_0.3.7.tar.gz Apr 19 16:33:15 hughsie_away or latest is: Apr 19 16:33:15 hughsie_away http://archive.ubuntu.com/ubuntu/pool/main/p= /powermanagement-interface/powermanagement-interface_0.3.11.tar.gz Apr 19 16:34:35 hughsie_away pjones, see how similar itis to pm-utils? Apr 19 16:36:20 pjones yeah, it does have a lot in common. Apr 19 16:36:48 pjones blah, all the scripts are separated into other pac= kages. Apr 19 16:37:00 * pjones downloads the whole distro so he can look at the= m. Apr 19 16:37:03 hughsie_away yup, lol Apr 19 16:37:05 mjg59 The scripts are in acpi-support Apr 19 16:37:19 mjg59 Which actually needs rationalising, but still Apr 19 16:37:24 hughsie_away mjg59, what;s your take on all this? Apr 19 16:37:41 mjg59 Having a standard in hal makes sense, and suse be d= amned Apr 19 16:38:06 mjg59 Having hal call userspace code that calls another d= aemon that runs some scripts that trigger the suspend is madness Apr 19 16:38:28 hughsie_away no comment, as I think my opinion is known.. Apr 19 16:38:45 pjones acpi-support scares me. Apr 19 16:38:52 hughsie_away LOL:-) Apr 19 16:39:07 pjones mostly because the first thing I looked at was pow= ermgmt-base's "acpi_available.c" Apr 19 16:39:25 pjones which is a prime candidate for a one line shell al= ias if there ever was one. Apr 19 16:40:01 pjones (also: fflush(stderr). rotfl.) Apr 19 16:40:17 hughsie_away anyway... Apr 19 16:40:40 hughsie_away mjg59, pjones has concerns with machines tha= t don't have hal thatstill want this infrastructure in place Apr 19 16:40:58 hughsie_away i.e. the automodule unload, and ledblink hoo= k code Apr 19 16:40:59 pjones I just don't see what we gain by using hal to do i= t. Apr 19 16:41:57 mjg59 powermgmt-base? Apr 19 16:42:01 hughsie_away the only way it could be outside of hal, if = hal was to "depend" on this set of scripts Apr 19 16:42:07 mjg59 Oh, argh, pain Apr 19 16:42:14 mjg59 That's the Debian standardisation attempt Apr 19 16:42:17 mjg59 They're a bit mad Apr 19 16:42:25 hughsie_away no comment! Apr 19 16:42:26 pjones mjg59: http://pastebin.com/669435 Apr 19 16:42:29 pjones that's more than a bit mad. Apr 19 16:43:04 hughsie_away hmm Apr 19 16:43:09 pjones it's 20 lines of amazingly bad C code to do "[ -x = foo ]" Apr 19 16:43:33 mjg59 I think claiming that that code is copyrightable is= quite special Apr 19 16:44:04 hughsie_away mjg59, assuming we can get some standardisat= ion on the hook scripts, be they in hal, or in a new dep of hal, I guess = ubuntu would buy-into that? Apr 19 16:44:16 mjg59 hughsie_away: Absolutely Apr 19 16:44:26 pjones I think being willing to put your name on that is = pretty special. Apr 19 16:44:34 mjg59 Also, bear in mind that this should work on pmu and= apm systems Apr 19 16:44:41 pjones mjg59: ok, so have you looked at our pm-utils ? Apr 19 16:44:44 mjg59 pjones: Yeah Apr 19 16:44:46 hughsie_away mjg59, sre, of course Apr 19 16:44:55 pjones ok, then we need to come to some middle ground ;) Apr 19 16:44:56 hughsie_away mjg59, what you make of it? Apr 19 16:45:12 hughsie_away i think a new module might be the most "pc" = way of doing this. Apr 19 16:45:19 mjg59 I'm not entirely sold on the "Do vbetool stuff base= d on the graphics chipset", because the requirements do end up varying a = great deal with specific hardware Apr 19 16:45:43 hughsie_away but that doesn't matter if we dont install t= he hook for that. Apr 19 16:45:43 pjones mostly based on the bios, but yes. Apr 19 16:46:08 mjg59 Yeah. The Intel BIOS seems a great deal more standa= rdised than the others Apr 19 16:46:13 pjones (and I'm totally ok with separating the hooks from= the infrastructure, or a package not including all the hooks we've got, = etc) Apr 19 16:46:20 hughsie_away i.e. get the infrastructure in place, then l= et the distros do thier own thing. Apr 19 16:46:30 hughsie_away pjones, sweet Apr 19 16:46:37 mjg59 vbetool stuff ought to be settable via fdi files, I= think Apr 19 16:46:50 hughsie_away mjg59, then you have a dep on hal Apr 19 16:46:55 mjg59 Yes Apr 19 16:47:21 hughsie_away and you rely on a suspend coming from a dbus= method, rather than a script Apr 19 16:47:23 pjones mjg59: ok, what about a dbus hook to tell HAL to d= o that? Apr 19 16:47:33 mjg59 hughsie_away: Indeed Apr 19 16:47:38 pjones i.e. you run the script and there's a hook that te= lls HAL to do video stuff? Apr 19 16:47:43 mjg59 pjones: Urgh. Apr 19 16:47:48 hughsie_away agreed Apr 19 16:47:49 mjg59 I guess, yeah. Apr 19 16:48:00 mjg59 "Please reinit the video" Apr 19 16:48:06 hughsie_away you could expose methods on the video_device= object Apr 19 16:48:14 pjones I absolutely _don't_ want HAL driving the suspend/= resume process. But I agree that video is the one place where HAL may we= ll have the best info. Apr 19 16:48:17 mjg59 Sounds good Apr 19 16:48:19 hughsie_away like we do for setlcdbrightness Apr 19 16:48:41 hughsie_away pjones, i agree with that split Apr 19 16:49:18 mjg59 Right now, the video stuff is the main point where = we actually care about hardware magic Apr 19 16:49:24 pjones ok, so if hal has the methods on the video device,= then it's trivial to copy the NM script and do the right call there. Apr 19 16:49:28 mjg59 So abstracting that into hal makes sense Apr 19 16:49:31 pjones right. Apr 19 16:49:31 hughsie_away yes Apr 19 16:49:42 mjg59 And the rest can live outside Apr 19 16:49:49 pjones the rest of things we're more concerned with "this= module sucks ass" or "make the light go blinky" Apr 19 16:50:00 hughsie_away yup. Now, in hal, or out of hal? Apr 19 16:50:08 pjones hrm? Apr 19 16:50:11 hughsie_away this is going to be a super small shell scri= pt Apr 19 16:50:15 pjones yep. Apr 19 16:50:18 hughsie_away if there's no hooks Apr 19 16:50:51 hughsie_away and a couple of commands for cmdline suspend= and hibernate Apr 19 16:50:54 pjones hughsie_away: it'll look almost exactly like pm-ut= ils' network manager script, minus the ntpd crackrock. Apr 19 16:51:04 hughsie_away pjones, yes Apr 19 16:51:25 * thoenig (n=3Dthoenig@charybdis-ext.suse.de) has joined = #hal Apr 19 16:51:41 hughsie_away mjg59, what do you think of pm-utils code? Apr 19 16:51:42 pjones Just as soon as hal has this functionality, I can = add it to pm-utils . Apr 19 16:51:54 * davidz back Apr 19 16:52:08 hughsie_away davidz, you need to read the last few pages.= .. :-) Apr 19 16:52:11 davidz I did Apr 19 16:52:17 hughsie_away what you think? Apr 19 16:52:26 mjg59 hughsie_away: Uh, I'm fine with it Apr 19 16:52:26 * You are now known as hughsie Apr 19 16:52:26 -NickServ- This nickname is owned by someone else Apr 19 16:52:26 -NickServ- If this is your nickname, type /msg NickServ I= DENTIFY Apr 19 16:52:32 davidz I think it sounds good Apr 19 16:52:43 davidz you guys should probably move it to the mailing li= st at some point Apr 19 16:53:07 pjones yeah, I'm going to try to get a mailing list set u= p for pm stuff that's _not_ "fedora-laptop-list" Apr 19 16:53:22 >NickServ< identify **** Apr 19 16:53:23 -NickServ- Password accepted - you are now recognized Apr 19 16:53:23 * services. sets mode +e hughsie Apr 19 16:53:23 -MemoServ- You have no new memos Apr 19 16:53:26 pjones (note that I got pm-utils moved to externally visi= ble cvs yesterday) Apr 19 16:53:26 mjg59 Yes Apr 19 16:54:38 dannyK mjg59: What do you mean with this ?: [17:36] Having a standard in hal makes sense, and suse be damned Apr 19 16:54:39 pjones Right now tarball releases live at http://people.r= edhat.com/pjones/pm-utils/ ,=20 Apr 19 16:54:52 dannyK why should suse be damned? Apr 19 16:55:15 pjones notting is bringing storage space for projects lik= e this up at the next fedora board meeting, so hopefully that'll move to = some place more sane. Apr 19 16:55:45 mjg59 dannyK: Because I think the mechanism that suse cur= rently uses (hal calls powersave which calls powersaved which triggers th= e suspend) is madness Apr 19 16:56:00 pjones yikes, that's way overcomplicated. Apr 19 16:56:37 mjg59 pjones: Do you do other laptop stuff, or just power= management? Apr 19 16:56:55 pjones mjg59: actually I'm the installer guy, but, well..= . Apr 19 16:57:00 mjg59 pjones: Ah :)=20 Apr 19 16:57:04 dannyK Yes and the hal/g-p-m way is better ? This way is = only used under gnome ... und KDE we use the nuic DBUS interface of power= save Apr 19 16:57:16 pjones (one of 5 installer guys, anyway) Apr 19 16:57:31 mjg59 dannyK: Given that hal has sufficient privileges, I= think it makes sense for hal to trigger the suspend rather than call out= to yet another daemon Apr 19 16:57:36 pjones mjg59: I'm somewhat freelance when there aren't ma= jor deadlines. I hack on what's in my way, mostly. Apr 19 16:57:56 hughsie pjones, the idea of having a resume.d, suspend.d = and hibernate.d irectory for files if a nice one, better than the pm-util= s way. Apr 19 16:57:59 mjg59 dannyK: Or, to put it differently: powersaved shoul= d provide policy, hal should provide mechanism Apr 19 16:58:23 dannyK this deamon exist much more longer than hal and he= do a really well job Apr 19 16:58:26 mjg59 pjones: We've been doing a lot of work on laptop ho= tkeys lately, and it would be good to find some means of sharing that wit= h other distributions Apr 19 16:58:42 dannyK Its not a reason do damn suse for this project Apr 19 16:59:00 mjg59 dannyK: I'm not saying it's bad code. I'm saying th= at, given that hal now exists, it makes sense for powersave to use hal (r= ather than vice-versa) Apr 19 16:59:33 pjones hughsie: ugh, that means pm-action has more logic = in it. Apr 19 16:59:39 pjones hughsie: that way lies madness Apr 19 16:59:48 pjones mjg59: definitely Apr 19 17:00:13 hughsie pjones, i don't see how, instead of one file hand= ing if action =3D=3D suspend, if action=3D=3Dresume Apr 19 17:00:19 mjg59 pjones: But right now there doesn't really seem to = be any way of doing cross-distro (but Linux specific) development Apr 19 17:00:21 hughsie we just split it up into 3 directories. Apr 19 17:00:27 pjones hughsie: also, I like having suspend+resume in the= same file. It means logic for the same conceptual thing sits at one pla= ce. Apr 19 17:00:38 pjones hughsie: split those up and you'll get more "we fo= rgot to hit the other file" bugs. Apr 19 17:00:40 dannyK And IMO it make sense to have a own daemon for thi= s ...but i don't plan to restart thus useless discussion ... I prefer the= suse solution .. and I also dont damn other distos for their way Apr 19 17:00:58 hughsie pjones, i guess... mjg59 what do you think? Apr 19 17:01:00 pjones mjg59: we should set something up a la freedesktop Apr 19 17:01:16 hughsie seems a bit trivial for that, but it seems the ri= ght place Apr 19 17:01:17 mjg59 I'm entirely unconcerned by what the script actuall= y looks like :) Apr 19 17:01:22 hughsie mjg59, lol Apr 19 17:01:48 mjg59 dannyK: No, I'm not saying that powersaved shouldn'= t exist. I'm saying that powersaved should decide what to do, and then as= k hal to do it Apr 19 17:02:04 dannyK mjg59: We have here enough problems with hal atm .= .. I don'T need one more task in HAL which make problems .. Apr 19 17:02:07 mjg59 That way powersaved doesn't have to be linux-specif= ic code Apr 19 17:02:22 dannyK why should i ask hAL if I also can do it by my sel= f Apr 19 17:02:29 pjones who cares if powersaved is linux specific code or = not? Apr 19 17:02:41 pjones (there are still other OSes? ;) Apr 19 17:02:44 mjg59 It's a matter of good design Apr 19 17:02:50 pjones that's a strawman. Apr 19 17:02:57 mjg59 Or, more practically: Apr 19 17:03:01 hughsie hmm, but i agree with mjg59=20 Apr 19 17:03:07 mjg59 The only thing that should care about the bizarre k= ernel interfaces under /proc is hal Apr 19 17:03:16 mjg59 That way we only have one thing to fix when they ev= entually move somewhere saner Apr 19 17:03:36 pjones "Now who's being naive, Kay?" Apr 19 17:03:56 hughsie bbiab, picking up g/f Apr 19 17:04:24 mjg59 (This is pretty much the opinion of the kernel pm g= uys, in any case) Apr 19 17:04:36 kay pjones: naive? Apr 19 17:05:19 pjones It'd be nice if everything that touches /proc inte= rfaces were at one place. I don't know what planet that'll ever happen o= n, but hey, sounds great ;) Apr 19 17:05:19 * [ervin] (n=3Dipsquad@lns-bzn-36-82-251-14-155.adsl.prox= ad.net) has joined #hal Apr 19 17:05:43 davidz kay: it's a quote Apr 19 17:05:46 davidz kay: from the goodfather Apr 19 17:05:48 mjg59 Well, as hal grows in functionality, we're graduall= y getting there Apr 19 17:05:57 * ervin has quit (Nick collision from services.) Apr 19 17:05:57 * [ervin] is now known as ervin Apr 19 17:06:01 davidz kay: where Michael Corleone's wife is named kay :-= ) Apr 19 17:06:07 mjg59 Then we just need to beat everyone until they use h= al Apr 19 17:06:10 davidz kay: at least that's what I thought pjones meant := -) Apr 19 17:06:18 kay davidz: oh, i see :) Apr 19 17:06:36 pjones davidz: It's a sad day when only one person catche= s a godfather reference. Apr 19 17:07:26 davidz pjones: well, kay's name is also kay :-) Apr 19 17:07:35 pjones All the more reason he should know it ;) Apr 19 17:07:48 davidz Michael: My father is no different than any powerf= ul man, any man with power, like a president or senator. Apr 19 17:07:48 davidz Kay Adams: Do you know how naive you sound, Michae= l? Presidents and senators don't have men killed. Apr 19 17:07:48 davidz Michael: Oh. Who's being naive, Kay?=20 Apr 19 17:07:50 davidz ... Apr 19 17:07:51 davidz hehe Apr 19 17:08:29 pjones It's such a wonderful scene. Apr 19 17:08:32 davidz totally Apr 19 17:09:19 pjones (it should be noted that Michael is wearing his ar= my uniform at the time) Apr 19 17:10:19 davidz pjones: btw, the SUN guys have been working on imp= lementing HAL on Solaris for quite some time Apr 19 17:10:29 davidz pjones: at first only storage stuff Apr 19 17:10:38 davidz pjones: but I don't see why they couldn't benefit = from this Apr 19 17:10:49 pjones Sure. I just don't care. Apr 19 17:10:53 davidz pjones: and, conversely, all the Linux distros can= benefit too Apr 19 17:11:09 davidz well, I care, because if we're all improving the s= ame thing it gets better Apr 19 17:11:16 * pjones got tired of trying to cooperate with Sun and ha= ving them make exactly no attempt sometime in the mid 1990s Apr 19 17:11:17 davidz instead of a lot of half-baked stuff Apr 19 17:12:06 mjg59 Sun throw some development at gnome, so it's occasi= onally nice to encourage the belief that we still think they have some re= levance on the desktop :) Apr 19 17:12:18 davidz section 508 stuff for example Apr 19 17:12:51 * pjones goes to lunch Apr 19 17:30:45 hughsie back Apr 19 17:30:50 hughsie so, a todo. Apr 19 17:31:02 davidz hughsie: perhaps it would be useful to write a sum= mary to the list? Apr 19 17:31:09 davidz enclosing the log as an attachement? Apr 19 17:31:13 davidz attachment even Apr 19 17:31:15 hughsie davidz, yes that would be best Apr 19 17:31:24 davidz cool Apr 19 17:32:59 hughsie davidz, how easy to get stuff hosted on freedeskt= op? Apr 19 17:35:22 * davidz has quit (Read error: 104 (Connection reset by p= eer)) Apr 19 17:35:25 * davidz_ (n=3Ddavidz@wireless-152.media.mit.edu) has joi= ned #hal Apr 19 17:35:37 * davidz_ is now known as davidz Apr 19 17:35:42 * Notify: davidz is online (clarke.freenode.net). Apr 19 17:35:51 kay hughsie, davidz: why not put the stuff into the hal c= vs root like policykit? Apr 19 17:35:59 davidz hughsie: sorry, I lost the network Apr 19 17:36:16 hughsie kay, thats a good idea Apr 19 17:36:30 hughsie davidz, can you make it so? Apr 19 17:36:31 davidz hughsie: hold on a second Apr 19 17:36:40 hughsie i'm hanging Apr 19 17:36:57 davidz hughsie: maybe it's premature to think about CVS b= efore we all decide external tarballs is the way to go Apr 19 17:37:04 davidz I think we need to move the discussion to the list Apr 19 17:37:15 davidz but, in any case, getting CVS on fd.o is easy Apr 19 17:37:20 hughsie okay, i'm summing up a list of points now Apr 19 17:37:25 davidz sounds good, thanks :-) Apr 19 17:39:11 hughsie what about a name? Apr 19 17:39:26 hughsie i would rather not use pm-utils or pmi.. Apr 19 17:39:43 davidz again.. a name, for what? Apr 19 17:39:51 davidz It's not clear to me we need an external package Apr 19 17:40:02 davidz (and, yea, naming is pretty difficult) Apr 19 17:41:12 pjones mjg59: so is vbetool post supposed to work on inte= l 955gm? Apr 19 17:41:41 davidz hughsie: (just to add... we probably want an exter= nal package.. but first things first) Apr 19 17:42:08 hughsie davidz, ohh i know, by my summary has "project x"= all the way thru so far Apr 19 17:42:27 hughsie i'm committing nothing don't worry! Apr 19 17:42:31 davidz haha Apr 19 17:42:34 davidz Project X Apr 19 17:42:35 davidz fun Apr 19 17:42:42 hughsie erm... no. :-) Apr 19 17:43:06 pjones why do we care about the name? Apr 19 17:44:13 hughsie pjones, just so i don't have to write project x a= nymore Apr 19 17:44:17 hughsie :-) Apr 19 17:44:55 * davidz Gentlemen, no fighting in the war room! Apr 19 17:44:57 davidz :-) Apr 19 17:45:35 * davidz is now known as davidz-lunch Apr 19 17:45:42 * Notify: davidz is offline (clarke.freenode.net). Apr 19 17:45:47 pjones hughsie: ok... what's wrong with pm-utils or pmi? Apr 19 17:45:56 hughsie i think power management tools is the best name. = pmt... :-) Apr 19 17:46:30 pjones (I would actually say that we should stick with pm= -utils if we're using that code, just because it makes things easier to t= race back if anybody tries to look for changelogs and whatnot) Apr 19 17:48:42 pjones (not to mention it means we can pretty much leave = CVS as it is right now, and just give people accounts/ACLs if they need t= hem) Apr 19 17:49:32 hughsie hmm... i think ultimatly this belongs outside @re= dhat, politically if you see what i mean Apr 19 17:50:11 hughsie and i think we need to split the infrastructure f= rom the scripts Apr 19 17:50:19 hughsie like ubuntu do with acpi-support Apr 19 17:50:59 pjones Will anybody really care that the cvsroot they're = using is @redhat.com? Apr 19 17:51:52 hughsie me, no, but it would look more independant if it = was f.d.o Apr 19 17:51:59 * pjones shrugs. Apr 19 17:52:07 hughsie politics i think Apr 19 17:52:12 pjones I can't say I really care; I think you're wasting = your effort, but go ahead. Apr 19 17:52:19 hughsie no, no plans for me Apr 19 17:52:38 hughsie i'll raise it as an option to hal-delel and see w= hat ppl say Apr 19 17:52:38 pjones Can we at least take a tarball of the cvsroot if w= e're do, so the logs stay Apr 19 17:52:48 hughsie ohh, yes, agreed Apr 19 17:53:22 pjones and actually, I don't think we need to separate th= e scripts from the infrastructure Apr 19 17:53:38 pjones It may be worthwhile for distros to make two packa= ges, but I think we want to leave the _sources_ together. Apr 19 17:54:40 hughsie ala have one tar.gz but 2 debs for instance? Apr 19 17:54:50 pjones (and really... two tarballs for 800 total lines of= content? ;) Apr 19 17:54:55 pjones right Apr 19 17:54:56 hughsie point taken! Apr 19 17:55:15 hughsie thanks for your help with this btw. Apr 19 17:55:21 pjones no problem Apr 19 17:55:24 pjones thanks for asking ;) Apr 19 17:59:53 hughsie pjones, you got that cvs url handy? Apr 19 18:00:19 pjones :pserver:anonymous@cvs.fedora.redhat.com:/cvs/deve= l , module is "pm-utils". Apr 19 18:00:49 pjones (for people who we want to give rw accounts, they = need to make a fedora extras account and then I can have bill add them to= the ACL) Apr 19 18:02:11 hughsie okay, nice one, thanks Apr 19 18:13:43 hughsie pjones, why the thaw/resume difference in nomencl= ature? Apr 19 18:14:14 pjones what do you mean? Apr 19 18:14:54 hughsie when is thaw !=3D resume for actions? Apr 19 18:15:12 pjones I don't know. Right now there's no case that we'r= e shipping, but I don't know that that'll always be true. Apr 19 18:15:35 pjones (after all, 2 months ago there wasn't a case where= hibernate varies vs suspend...) Apr 19 18:15:39 hughsie i guess. Apr 19 18:16:28 pjones I try not to make plans that require me to know ev= erything everybody else will do in advance, if I can help it. Apr 19 18:16:56 hughsie sure, gotcha Apr 19 18:17:01 hughsie is /etc/sysconfig a rh thing? Apr 19 18:17:05 pjones yes. Apr 19 18:17:36 pjones We can change that to something defined at build t= ime if you like... Apr 19 18:17:44 hughsie i think we have to Apr 19 18:17:47 pjones nod. Apr 19 18:17:58 pjones <-- happily accepting patches ;) Apr 19 18:18:17 hughsie okay, just getting an idea of the detail Apr 19 18:19:11 hughsie pm-utils is actually quite sane.. Apr 19 18:19:40 pjones Thanks ;) Apr 19 18:19:44 hughsie echo "export ${1}_STATE=3D$2" >> /var/run/pm-susp= end Apr 19 18:19:50 hughsie why that file? Apr 19 18:19:52 hughsie seems odd Apr 19 18:19:58 hughsie /var/run/pm Apr 19 18:20:05 hughsie /var/run/pm-state Apr 19 18:20:17 pjones there was another script that was using /var/run/p= m when I typed that, I think... Apr 19 18:20:25 pjones pm-state makes more sense, I guess. Apr 19 18:21:08 pjones (not to mention "pm" is a little sparse of a name = for that sort of thing) Apr 19 18:21:21 hughsie sure Apr 19 18:21:24 pjones actually, pm-suspend is pretty good -- you know wh= ich "pm" we're talking about from reading it. Apr 19 18:21:38 hughsie but we also use it for pm-hibernate Apr 19 18:21:48 hughsie Also why the checks for /.suspended ? Apr 19 18:21:49 pjones (/var/run/pm might be package management, for exam= ple) Apr 19 18:22:01 pjones because acpid will give you duplicate events in ra= pid succession sometimes :/ Apr 19 18:22:15 pjones (which I think is a BIOS bug, but I'm not 100% sur= e) Apr 19 18:22:28 hughsie this shouldn't be handled here tho, surely? Apr 19 18:22:35 hughsie hal layer? Apr 19 18:22:50 pjones what winds up happening is that you get the suspen= d event right as you start resuming Apr 19 18:22:59 pjones and you go back to sleep. Apr 19 18:23:09 hughsie ohh i see Apr 19 18:23:11 hughsie gotcha Apr 19 18:23:13 pjones Why not there? It's easy to do and it works. Apr 19 18:23:24 * davidz-lunch is now known as davidz Apr 19 18:23:28 * Notify: davidz is online (clarke.freenode.net). Apr 19 18:23:33 pjones the only truly ugly thing (ignoring video) that we= 've got in there right now is the ntpd stuff. Apr 19 18:23:46 hughsie video can go -> hal Apr 19 18:23:49 pjones But there's no way to make it better except to act= ually make ntpd not suck ass. Apr 19 18:24:29 pjones (i.e. make it watch the network through dbus) Apr 19 18:25:20 hughsie hm.. Apr 19 18:25:28 pjones when you stumble accross the ntpd stuff in there y= ou'll see what I mean ;) Apr 19 18:25:36 hughsie okay, i'm looking now --=-VRQkX4i+JbK6gbeRQ2gn Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline --=-VRQkX4i+JbK6gbeRQ2gn-- --------------010900020203080700080307--