[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
SEUL-Leaders: Aug 05 IRC log
**** Logging Started : Tue Aug 05 19:49:35 PDT 1997
*** omega has set the topic: "Simple End-User Linux - technical meeting at 9pm PST"
*** omega has set the topic: Simple End-User Linux - technical meeting at 9pm PST
*** luka (~luka@PAPRIKA.MIT.EDU) has joined channel #seul
> yo
<luka> poing
* omega no, it's you ping, I pong.. ;-)
<luka> so, first thing:
> go ahead and type, I'll brb with some caffeine... ;-)
<luka> (hey, the pong came first, so i just lumped everything)
> (true)
<luka> ok. i'm sort of envisioning seul "extensions" rather than a full distribution, to start off with, at least. i think that with a prereq of a decent package manager, and other std stuff, we can in the long run "seul-ify" any distribution. but i think we will start our first impl with redhat.
<luka> so, specific extensions:
<luka> 1. replace the installer.
> this is the pre-1.0, I assume... 1.0 will be a near full-distrib
<luka> 2. seul db - sits beside the main package manager, and keeps track of all sorts of things (you know what i'm getting at)
> "registry"
<luka> 3. seul gui. standardized, friendly, and integrates with seul-db ((2) above)
<luka> 4. help system. integrates with other stuff.
<luka> 5. seul apps. includes base stuff direct from us, plus "seul-approved" apps, which all need to integrate to some extent with the rest of our stuff.
<luka> to pull this off, we'll need some glue in the form of an interface lib to go between the gui, the db, and everything else, mainly.
> the most interesting parts are the installer and the gui.... they take the most research.... everything else is "just coding" :)
> which reminds me, I'll be helping a coworker install Linux on their new (right from the mail) box soon, I'll be taking copious notes...
*** Virtue (~newbie@ppp-annex-0549.tor.total.net) has joined channel #seul
<luka> gui: xwindows does some cool shit (tm) just within the x protocol. i think we can exploit a lot of that in the gui interfacing.
<luka> what distrib you installing? redhat?
> right... the question is what to use... we use KDE, we're stuck with that, which includes the help system, etc...
> redhat 4.2
> but it comes preinstalled with dozeNT 4.0
> he's an engineer, but more the "gimme the tools and let me code" type
> not the tinkerer
<luka> ok. well, let me know how it turns out - especially anything he complains about :)
> right...
> I'll try to let him do most of it himself...
> the trick is going to be FIPSing the disk... :(
> FIPS doesn't do NTFS, IIRC
> we may end up having to do evil things like 0 the disk and reinstall NT from CD, then do Linux...
> back to seul... the idea is to start with the bare essentials, then let them take over the entire distrib for 1.0?
> i.e. 0.xx contains the components, not a complete distrib
> actually that's more of an artifact of how the development will proceed, more than actual goals...
<luka> so on the gui side - i want to exploit features of x, especially the window manager abstraction (the whole "hinting" thing will finally get a lot of use - and we'll see how many developers tried to obey icccm at all :), also xrdb will be very useful.
*** Signoff: Virtue (Homer was here!)
*** Reconnected as omega at Tue Aug 05 20:18:28 PDT 1997
> ugh
*** Mode change "+o omega" on channel #seul by luka
*** Mode change "+i" on channel #seul by omega
*** Mode change "+p" on channel #seul by omega
<luka> what's up with that lossage?
> dunno
<luka> random irc death?
> apparently
> on the X front... there's the KDE that a few people want to use, but it does *everything* for you, in one nicely packaged system
<luka> i've noticed ircii goes down in a very unfriendly way when it loses server contact, but that's all the trouble i've had...
<luka> kde licensing seems to bite, though. thoughts?
> actually, it's Qt's licensing
<luka> yes.
> don't think we could use it unless Qt goes GPL
<luka> or we port it to something other than Qt, but i think that would be ludicrous.
> not something we want to do
> so KDE'
> KDE's out for the most part, for that reason and others
<luka> wrt that topic, here's the way i view gui things:
> I'd like to be able to provide the user with a list of window managers, desktop entities, etc., and have their apps "automagically" draw themselves (via hinting) correctly...
<luka> right. that is the effect that would be really good, if we can make it work.
> that involves getting the WM's to code to a pseudo-API, as well as apps, etc...
> pseudo in the sense that they have to comply with certain standard practices, not to mention IC3M
<luka> right. so the desired thing is a common gui interface that sits between the apps and the window manager, and makes things behave. we can control the wm's themselves to some extent by writing resource files for all the supported ones, but with this layer, we can more generally modify hints going to the wm's... might be useful, or might be more than we care to bite off...
> then there's the config issue...
<luka> but let me step back for a moment:
> every wm and app has a different config format, which means we need some kind of abstraction layer...
> go:
<luka> right. the technical side gets pretty heinous fast. so here's the Big Picture:
<luka> the effect we want to achieve is a uniform gui, across all supported wm's and supported apps, and we want to maximize the size of the supported software (within reason), while minimizing the effort that goes into us maintiaining things. (more:)
<luka> we can look at the uniform gui at two levels though... intra-app, and inter-app. this is something to keep in mind. (more:)
<luka> also, we have a Bare Minumum to meet. for any wm running - seul base apps (admin, etc) _must work_. beyond that, i think that we can restrict support to a few wm's that are good, as long as people who insist on oddball wm's can still run the really base stuff.
<luka> i see two basic markets for seul:
<luka> 1. people who want the friendly os. (think pretty gui - easy to grok the controls)
<luka> 2. people who want the utilities. (don't care about the gui, might run nawm (like me), but would find a highly automated, quick-installing, pretty much self-maintiaining linux system very useful (perhaps because they maintain 500 of them))
<luka> .
> right... the trick is to find wm's and apps that can either be "merged" into the seul effort, or will be willing to mod their code to meet up with us....
<luka> logging?
> logging this? yes.
<luka> good. gimme transcript when we're done. :)
> there's a hybrid of those two...
> there's the case where you have the IS person installing 500 of them, but you have to have the pretty gui for the actual user.
> that makes things interesting..:
> you have the IS person who likes to keep things nice and stable and standard, and you have the user who wants to tweak things, install their own editor that they like better than the standard one, etc...
> however, I don't think we want to be supporting big IS-controlled installations quite yet...
> let's get the GUI, installer, apps, etc., working before we go into that...
> :)
<luka> yeah - i didn't mean to imply distinct markets (prob the wrong word). think of dimensions, with jruser being some linear combination of them.
> right
<luka> so, on the is tangent - that triggers another topic.
<luka> lemme spew:
<luka> so, we've got all this talk about, "hey, there should be a config for foo" and "there should be a config for bar" (schools, homes, business, etc).
<luka> followed by debate about whether each is a noble cause that we should spend time on or whether it would flop, and make seul a waste of time.
<luka> so let's pull that out of the seul picture, if possible. let's try to design the system s.t. a configuration can be _programmed_ on top of a general seul system that supports lots of customized "typical" installs. (more:)
<luka> gains:
<luka> 1. schools/home/business/marine-science debate. we just win. the basic seul supports everything with minimal extra effort. so if one is a flop, seul is not a flop. i feel that just supporting the thing we want to directly (2 or 3) would already be as much or more effort than building the generalized system. so it is worth it if that is the case on closer analysis.
> but how do you propose to deal with that? separate distribs for things? or a question in the installer: "Are you a marine biologist?" :-)
> the actual core of being able to run under these different environments is really the easy part, in the sense that all you really need to do is make sure the apps are there for the users you are targetting, and make sure they all work together in whatever combinations are possible...
<luka> 2. Foonet, inc. wants to create a linux installation that they can deploy on 500 workstations with one click, and maybe a few additional clicks if a particular system has really evil hardware. They take seul-lib, write a config for the way they want their systems to look, click compile :), and have their own super-easy installation. for people who want this, the burden is taken off of us (it goes into their effort in specifying the config as well as they care
<luka> the seul config lib makes it much easier than creating their own boot/install system from scratch. (athena has done something sorta like that and it is still a bit flaky)
> in that sense you'd simply have a very configurable network installer, such that you have a boot/root disk with a config script for the installer
> the IS person can dump a floppy or 500 in the drive, 'burn' then with the install disk, then go to each machine and install them.
<luka> (mit would use it, for starters. just occurred to me:) - so would many other schools with sophisticated infrastructures)
> that's really all in the domain of the installer
> then things like upgrades go into the interface with the Linnet backend, for instance, which puts it in the admin area
> in the case of IS doing a mass install, we can provide a utility that goes and creates the boot/root, including their install config data, then dumps it to a dd image
<luka> that is what i was thinking (as a natural extension once we are doing that ourselves anyway)
> the real issue comes when IS isn't installing things for you...
<luka> hmm?
> when the end user is talking to the base installer program directly, how do we deal with this?
> do we act like w95 and give them categories?
> there are two choices, really:
> 1) installer does the basics, including the core system, a wm or two, a few applets, and the basic "office" suite
> 2) installer can install *everything*, including the obscure fish-ecology :) apps
> 2) is the harder one because that implies that the installer has the infrastructure (interface-wise) to deal with everything
<luka> what context is this in? who/where is the user doing this install?
> 2) is the harder one because that implies that the installer has the infrastructure (interface-wise) to deal with everything
> (up-arrow... :)
> home user, for instance
> basically, we've defined two types of installs - the mass IS-ish install, and the "home-user" install, which includes everyone else
<luka> home user would not see a marine-biology option - that would be added just for the guys at Foo-UnderwaterLabs, Inc to give to their employees.
> but the question is do we want to define an install type for every obscure profession, or just categorize apps enough so they can make intelligent choices and install time and beyond/
<luka> i think the options that will pop up on a Standard seul installer screen will be "1) typical install, 2)minimum, 3)everything, 4) custom". Typical is then followed by "are you a 1)home user, 2)business user, [hidden extras 3:run custom install for Foo, Inc]
> but do we want to play 20-questions with normal users?
> are you a parent? how many kids do you have? what ages? (for instance)
<luka> and then 1 and 2 (home and business) would go on to interrogate further about exact apps to install, if necessary, or just say Click here to install. for the average user - 2 clicks installs the whole system.
> "interrogate" is exactly what scares people away
> (btw, I question the use of the minimum install... great for trying things out, if you don't want to do anything, but for most people who'd be interested in a minimum install it's because they can't fit everything, thus they'd prob use 'custom')
<luka> are you a parent? how many kids do you have?, etc... may be way too damn roundabout. i don't think that level of psychoananlysis is necessary. i think "click here to install" with a well-thought-out typical home user install will work for 99%
> exactly
> not question people between home and business...
> or if so, make that question #1
> this all hinges upon the possibility that the home and business installs are really that different
<luka> 1> typical home install, 2>typical business install, 3, 4.
<luka> and the reason we win:
> they both want a word processor, they both want some kind of spreadsheet, they both want the nifty CD player applet, etc.
<luka> because that decision can be made at the last minute anyway, or at least after we have done more research through interviews, etc. because if we use the general install-config abstraction in our code, we can simply plug in or remove configs easily.
> right
> I don't argue the need for a generalized install/config thing
<luka> i think the generalized installer has definite merit if feasible.
> maybe we should just leave it at that for now, figure out the install types *much* later... :)
<luka> (but that is what needs to be specified and built :) yup. let's design the generalized installer abstraction, and keep, say, home/business/school (i.e. something familiar) in mind while doing so.
> right
> not tie ourselves into anything specific, for the reasons mentioned way up there ^
<luka> keeping them in mind will keep us from going off track, but remember that that is not what is being implemented immediately. also:
<luka> i think for our actaul coding strategy, we should implement the general installer, plus one very rough specific config just for test. it is very satidfying for programmer to write code that does something the first time.
> right... basically it'll be a custom install
<luka> and then go on to refine the general stuff, and once it is done, then fill in the specific configs for real:
> the config that is, it'll give us the full custom install menu/buttons
<luka> but it is important to keep an eye on development, and make sure coders are focusing on the generalized code being good, at least initially, rather than putting a lot of effort into a particular config.
> right, hence the initial completely custom install, so we *can* try all permuations
<luka> because the particular [test] config will probably get scrapped after we've had time to do research.
<luka> yup
<luka> sounds like a Plan. :)
> Da Plan...
<luka> that's what they say... yeah. er, that's what i say, actually. :)
> side note, go to http://www.rasterman.com/
> dr.11 of enlightenment released, try it out!
<luka> seen it.
> dr.11?
<luka> no - saw the page. will obtain the software soon.
> the one problem with it is that it isn't modular yet
> I would use it as my main wm if it were, so I could install an fvwm-like pager instead of the default
> 800x600 only goes so far...
<luka> brb
> I squeeze everything I can out of my screen, but enlightenment seems to be designed for 1280x1024++
> ditto on the brb, better wash some clothes...
<luka> back
<luka> which brings to mind the performance/hardware req thing. this is actually something that goes into system specs. (although i don't think it ever made it into the windows system specs, or something... we do not want /that/ to happen...)
<luka> we'll need to define some specs for what sort of performance/hardware-tolerance criteria seul needs to meet before we'll be willing to do a release.
> back
> the primary goal is to be better than windows... ;-)
> not hard, with the existing app base...
<luka> damn straight.
<luka> and try not to make linux crawl all of a sudden. old linux users should not ask "what the hell happened" upon seeing a drastic slowdown when they install seul.
> right
> the trick is to minimize what happens during normal ops
> as with X, define protocol, not implementation
> er...
> implementation in the sense of putting a real layer of stuff between this and that
<luka> basically, i think some performance hit is to be expected, but we should try to minimize it. (and i think the word "try" should be in the official spec. (along with some precise upper bounds, too))
> and some upper case...
> */TRY/*
> as far as min hardware, I say we should assume 8MB minimum, 486dx33 minimum, though the combo is painful...
<luka> i'm not sure about that yet. i'm still thinking about the exact min hardware issue.
> you want to go lower or higher?
<luka> can you make an argument for me for that config (other than the "most users" argument)
> you mean the 486dx33?
<luka> yeah
<luka> and 8m
> well, there's really no argument....
> it's just that it'd be nice if we could support it
> without a drastic slowdown
> for the most part:
> I think we should assume that the normal user for seul will be installing on a <2yr old computer, at least a P5/100 with 16MB of RAM
> 8MB sometimes, hence the ability to work on such a system
> it's mostly for people with older hardware that would want to be able to run on a 486dx33
> for instance, what if Granny wanted to do e-mail, nothing more...?
<luka> i think its reasonable to say that 386's are getting tossed out the windows of tall buildings (quite literally in some cases), am i right?
> righty
> you're not gonna buy Granny a P5/233MMX for e-mail, are you?
<luka> but there are still plenty of 486's in use.
> right
<luka> i think that may have to do with the rise in computer buying right around 486 time, too.
> so you put together a 486 with 8MB of RAM and install seul
<luka> and 8m is quite typical.
> thus we should be able to support it, if not fast, but at least run without paging to hell and back
> remember, with linux you can run on hardware that Windows <3.0 had a hard time on...
<luka> (i have one precisely like that sitting here, in fact. and friend of mine was actually still using this for everything half a year ago (till i helped her replace it with a 686))
> our goal should be to facilitate the re-use of old hardware
> I have a box with 486dx33/8MB/211MB that I could donate to a school if they were to run Linux on it
<luka> so, yeah. back to 486/8m should be done. i think 386 is allowed to bite the dust quite a bit though...
> that is, if I weren't using it for a dev system
> :)
> 386 - no go
> or "Nova"
> though I have a 386sx16/2MB that I *could* run a heavily tweaked version of Linux on... :)
> er.. it has 6MB now... nevermind
<luka> the 386 should probably run seul, sort of, in a minimal sort of mini-workstation/terminal configuration, and not suck more than the 386 inherently does. but beyond that, heavy-duty apps should not have to work on 386.
> right
> we can advert as "use your old 386 computers again - as terminals!"
> and you can even do some minimal processing on them if you *have* to
> network terminals, that is...
<luka> yeah, whatever. that sounds like the right track to be thinking along though. right?
> yup
<luka> so, um - finish up that html on the group structure -- we have yet to organize our team, remember?
> random:
> "It is by caffeine alone I set my mind in motion."
> right
<luka> i've seen that somewhere.
> theme o' my life... ;-)
<luka> well, i've got this jolt bottle, which i reuse for water now. fortunately, i have not had a need for the original contents in several months. but i suspect i'll be back to the old routine as soon as september rolls around...
> oh yeah...
> reminds me...
> september is something we're gonna have to plan carefully for...
> make that sept-june...
<luka> yup - how do you expect your load to be?
> not too bad, as I have to keep up at work too...
> I'm on the 5-6yr plan :)
<luka> i've determined i'm *not* on a 3-year plan. :)
> you sure?
> :)
> just seems like it somethines, eh?
<luka> well, theoretically i could.... uh.. no.
> no...
> Bad Thing
> maybe we should formalize "Bad Thing" and "Good Thing" and use them in our documents... ;-)
> (and trademark them... :)
<luka> i think Bad Thing almost completely summarizes that idea.
<luka> i think they're already Trademarked(tm)
* omega ducks to avoid the shotgun shell
> whew, that was close... ;-)
> 110% GPL
<luka> checking zephyr logs...
<luka> 132288:right, lynx should get The Right Thing (tm) for each platform..
> 132288?
<luka> 254347:netdate probably does The Right Thing(tm) and sets your
<luka> (line num)
> from what file?
<luka> 495047:Bad Thing (tm).
<luka> there it is.
<luka> zephyr help log. good one to start with...
> ah
> maybe we should search Gutenburg for them... ;-)
> wonder how many matches we'd get with that capitaliztion...?
<luka> so, what's your free time outlook for september+ ?
> prob 9hrs school over 2 days/wk, work 3days, weekends "free"
> hr is loosely defined... ~50-55min
<luka> that's sounds quite lax - sort of a half-coop type deal?
> coop? nope, though ~=
<luka> yeah, that is what i thought...
> SQNT job is independent of school
> I've been working there for >1.5yrs now, and I'm not even a sophomore in college yet ;-)
<luka> but they let you cut down school hours to double with work, or something?
> you mean cut down work hours to go to school - yes
<luka> ah, i see... :)
> I go to 3x7/wk during school
> er, 7x3...
<luka> i've been fattening myself up on cash this summer. i think i will cut work hours to about 0 starting september, and try to coast through til next summer on just an insane load of school work. it shouldn't be that bad though; i'm used to working > full time during term in the past...
> ouch
<luka> it's sick how i'm looking forward to this term, even though i'm expecting to take 2 Death (tm) classes quite possibly...
> sounds fun...
<luka> or something :)
> hmmm... well, back to seul?
<luka> ok. html structure doc? can you finish it now? let's get it out of the way finally.
> ok, this is the conversion of the fig + other stuff?
<luka> yeah
> other stuff == ?
<luka> fig contains seul-<toplevel>, other stuff is the mail i sent containing seul-dev-<foo>
> have you rewritten that at all?
<luka> not really, though if you hand me an html outline, i can probably fill in the whole thing really fast. that may be the best way to proceed.
> ok, lemme get that done really quick (~5min)
<luka> eek. it's 1 am - i didn't notice. but we did get quite a bit productively discussed... if you can pass that html outline off to me, i don't have to be up too early tomorrow, so i can probably fill it out tonight.
> working on it now
<luka> cool. i will take a little break then, and await that...
> ok
<luka> in fact, i will be back in 30 minutes... send email if you have to leave before then; otherwise i'll come back to irc.
> I have the skel
<luka> oh. ok. cran?
> lots of "para<p>" for you to fill in...
> transferring
<luka> works for me. thanks.
> ~/Public/dev.sdoc (html really...)
> I think I'll take my laptop, find a more comfortable position, and start hacking out random thoughts on "vision" stuff...
> maybe meet back here in ~35min?
<luka> ok. if you email me a semi-organized list-o-stuff, i can respond with organized comments. yeah - i'll be back in 35 min or so.
<luka> talk me at cran to get my attention.
> just stay here
> leave a window up...
<luka> works for me
> cya ~32min
> I've got an analogy for ya: Linux is Legos - lots of parts that you can do anything with
> existing distributions are like the starter sets, or those that give you a large selection of parts but nothing specific to build
> SEUL is like a large model set, with lots of alternate models you can build
> .
<luka> good one. sounds like something we want others to hear. :)
> right... thought of it while wasting a few cycles on my latest project - a robot leg like that used on the JPL 'bots like Genghis
> oh, did I mention that Linux is like Technic Legos... :)
> not the regular ones, where you build fire engines and stuff that are about 1.5" wide...
<luka> though it's not hard to take an analogy too far... ;)
<luka> but, yeah. seul builds big things easily.
> our job is to provide the plans for the big things, and maybe put them together while we're at it... ;-)
<luka> mebbe i should get in touch with some lego employees - see if i can get some reusable tips out of them. ;) but i suspect it will be similar to the usual successful system-product design methods.
> hehehe
> btw, my spew is in cran:/home/omega/Public/random
> 5831bytes
> coredump, actually
> hehe.... I kinda stopped in the middle of a sentence... ;-)
> more random? maybe later, when my brain's cooled a little... ;-)
<luka> have you used win nt much?
> just as a user with Citrix, the thing that does doze over X
> I know, nt is multiuser, but not as good as Unix
> in 'random' multiuser is defined as capable of dealing with multiple users one at a time on console, which even w95 can't do
<luka> nt actually deals somewhat well with multiuser config's. (as opposed to win95, which just plain doesn't)
> haven't gotten to the true multiuser stuff yet
> er, ditto again... ;-)
<luka> of course, no ms product deals well with multiple simultaneous users, especially on console. and i haven't even managed to get nt to put its console on a separate display or anything remotely useful along those lines.
> ooh, idea: single machine, up to 4 Mystiques, and USB keyboard and mice, you could truly run 4 people off one PC... ;-)
> NT server can, kinda
> we build them at SQNT, up to 32 procs, though that performs as well as about 0.5 procs... ;-)
<luka> kinda?
> can do multiple users, that is
> Citrix is one example
<luka> elaborate on "kinda" please...
> ;-)
> NT server as we sell it is mostly for SQL stuff
> but with Citrix you can:
> well, each user can log into the machine with a 'r[e]sh machine something -display whatever:0'
> it pops up a window via X that gives you a login as if you were on the console of the machine
> you have very little control over you environ, of course, it's all IS controlled
> actually, it's almost the same as having an NT workstation on your desk that logs into an NT domain server, with the server PM icons and user PM icons
> what was really funky was I was using and NT workstation for a while and then used Citrix and got the same environ, +-
> cause we have a large disitributed NT server cloud
> .
<luka> sort of the opposite of my work setup - i run eXceed on my nt box, and xhost a linux machine to do my work, because they want us running nt on our machines :(
> that's what most people do
> in engineering, that is
> almost everyone sits in front of a w95 box running eXceed talking to the Unix servers
> it's the rest of the company, i.e. not engineering, that uses the 95/NT stuff
> basically, engineering uses PCs as X terminals, with the added "feature" of being able to use MS-Mail
> I'm one of a very small minority that still uses Xterminals... :(
> s/:(/:)/
<luka> which consists primarily of ms exchange, word, and powerpoint... man, if we could just have perfect equivalents for those three apps under seul...
> right
> that's why I'd like to get user-oriented teams that focus on one type of user and push in that dir
<luka> we just switch from smtp to exchange at ice. bites.
<luka> is exchange protocol documented officially somewhere?
> one team for engineering people, and we could easily provide an alternative for 3/4 of SQNT engineering to switch to
> exchange servers can do pop/imap :)
> post 4.0.0, that is
> 4.0sp2 or something and above can do it
*** CTCP Reply from luka: PING - 5 secs
<luka> cool.. i think i may hack up a linux mail client at work soon. :)
> hehehe
> that's why we can target people like SQNT pd-eng with a lot of success...
> you know.... ;-)
> idea:
> the test engineer just left our group (making me Mr. Test), which leaves a spare computer... ;-)
> P5/166, 32MB, 21", 2.1G, etc
> I could convince our new manager to let me set up a test station for people to maybe switch to Linux with...
> we could save IS costs that way ;:)
> s/;//
> each network feed for a PC costs us $700/mo :(
> I run my system behind a dual P5/90 server, with private coax,we could run the entire group off that, assuming we can set up the server as a full TeamNet gateway withreal IP's behind it...
<luka> what
<luka> what
> save ~$3k/mo
<luka> erg
> ?
<luka> what's the "notebook-tag idea" in windows control panel?
> load up the "system" thing in w95
> there's a tag for performance, a tag for this, that,e tc.
> s/tag/tab/
<luka> i'd rather not, thank you. :) but, yeah, i remember now...
<luka> tabs. right?
> did I mispell it?
> oops...
> it has it's advantages, namly that it's physically based on real things that people use
<luka> yeah. but actaully, tabs are considered one of the great ui innovations of our times by the "experts", from what i know.
> but this is where we start reading the "experts" opinions
> exactly
<luka> i've seen tabs complemented a lot by ui experts, and never commented badly upon.
> prob a reason for that..
<luka> otoh, the best thing to do if you want a ui expert to punch you in the nose, is to use a physically-based gui element (or any other element) that is not necessary in a computer context. eg, notebook spirals. :) clutter is bad. tabs fight clutter.
> right
> spirals are baaad
<luka> tabs also align well, inherently. alignment is a good thing.
> nice and compact/efficient
<luka> yup. hence the incredible praise.
> you agree with me on the filesystem-like hierarchy idea for config?
> unfortunately, that's what linuxconf uses
> dunno about figurine
<luka> fs-hier: where is the reference?
> right below the tab thing
<luka> now that one is hot for debate. i hear the mac layout is the better motif, though i personally can't stand the mac layout (nor the "my <foo>" crap win95 added, and deal best with win3.1-style strict hierarchy (or just a shell prompt, which i suppose i map easily to the win3.1 style fs gui)
> are we one the same page?
<luka> i'll definitely have to get many second-opinions on that topic...
<luka> the filesystem-like tree is
<luka> not good for avg users. it's confusing, and hard to properly organize
<luka> things for avg users to find. sure, it matches the underlying structure,
<luka> but that's exactly what we're trying to hide, remember?
> I'm talking about user and system-level config in a format like the the doze exploder filesystem view, with the + and - stuff
> how is mac layout related?
> not the filesystem at all, just filesystem-like in that you have <foo> with a plus next to it, which expands to show <bar> underneath and indented, as a "sub-option" to <foo>
<luka> mac layout uses folders which expand into windows, that contain the contents of that folder (==directory).
> for the filesyste, yes, I'm talking about the configuration pages
> the windows control panel uses the mac style somewhat, as you have icons that open windows for each config thing (mouse, keyboard, date/time,e tc)
<luka> oh, yeah... yuck... are you talking about things like the "devices" control panel?
> yeah
> used for general config, though, not just devices
<luka> excuse me while i projectile vomit
> was trying to define it by example...
> hehehehehhee
* omega ducks
> geez, you broke mach with that one!
<luka> yeah... i think that's just really the wrong format for config options...
> it's the wrong format for anything but the fs, which is designed that way
<luka> right.
> so that's an 'agree'
<luka> yup
> any other comments on the rest of 'random'?
* omega will brb...
<luka> i think the login completion is a bad idea.
* omega is back
> true.. just an idea
> but the red/green light?
<luka> ok
<luka> re gui, again:
<luka> we want two things - a good interface, and a "transition" interface (for various different users)
<luka> :
<luka> i'm thinking it is important for us to have a standard "seul look", that just really works well, and that should be the main targeted interface. if it's well-designed, it should speak for itself, and users should not need the compatibility crutch. however, we don't want to throw away the great support x has for personalization and different wm's, and if we support all that, then adding compat support for, say win95, mac, and ...amiga? should happen as well.
> compat = transitionary, right?
> unless they like it and stick with it
<luka> yeah.
> I know Jis is working on getting his box to look like a Mac (ew!)
> some of it's good, like the menu at the top
> or more precisely, the button group in the top-right
<luka> but think of how many users used to just run the default fvwm with steelblue background, perhaps adding a bg pixmap, but seldom switching wm's... many people would gladly stick with "seulwm" so to speak, especially if it is cool.
> right
<luka> /run/ran
> I think getting enl configured carefully would be cool
> especially if it goes modular so people can add paging and so on if they want it
<luka> i think enl would hose the hell out of a 486 :)
> true
> enl and fvwm as our two primary wms?
> for instance
<luka> otoh, i think fvwm is highly standard, supported, configurable, and generally hackable (just look at fvwm95)
> fvwm95 is a heavy code hack, not config
<luka> yes - that was definitely in the "hack" category
> as long as you realize that it's a code hack and not config :)
<luka> yes. i do. :)
> good, just checking
<luka> you don't find "pop-up taskbar, with icon-storage" in the fvwmrc :)
> true
> not yet, at least
<luka> hehe. though there are several win95-ities i would hope never make it into a typical conf options...
> such as?
<luka> such as the #$!@$#$%%$!#&^&^@$ login dialog and associated startup stuff, especially the ctl-alt-del thing. and the damn "click here to begin" that just won't go away...
> tweakui can kill the click here thing, and the login thing is just plain broke
<luka> "just plain broke" may actually fit right in that character string: #$!@$#$%%$!#&^&^@$
> exactly, in fact
> just plain broke
> #$!@$#$%%$!#&^&^@$
<luka> (with the quotes)
> oh, two chars off, darn
> a decent login implementation under SEUL will be something to show of, most definitely...
<luka> anyway, rest of rambling looks like good rambling... i wanna go now. i will take care of the group breakout from here, and give you stuff to put on the web soon. how quickly can you get a web update done, to add this group-layout stuff after i have it ready?
> once I check it over and check it in, I can get it on the web in <2min
<luka> cool. when will you be available tomorrow (ie today)?
> er...
> if you have something, e-mail me at omega@sequent.com and I can take care of it
> if I'm not hacking on some problem or another that's a redlight ;-)
> I'll keep a window on cran, too
<luka> ok. will do. i'll probably be in touch mainly via email for a little while anyway - my schedule is about to get historically messy... be warned, i will probably not show up at irc regularly for a little while.
> work?
> s/historically/histerically/?
<luka> i think either works, but my spelling is the more accurate term. yeah, work schedule. trying to fit everything i do in 24 hours/day, and get some regular rest too...
> you only have 24?
<luka> i know, i know...
> hehehe
<luka> well, l8r...
> well, keep the e-mail somewhat verbose and all will be fine
<luka> yup. bye.
> cya
*** Signoff: luka (Leaving)
Erik Walthinsen - Programmer, webmaster, 3D artist, etc. __
__ / /\
/ \ omega@sequent.com Work: (503)578-5314 / / \
| | M E G A omega@aracnet.com Home: (503)281-4281 / / /\ \
_\ /_ psu12113@odin.cc.pdx.edu Majoring in CS / / /\ \ \
/ /_/__\ \ \
Omega Station: http://www.aracnet.com/~omega/ /________\ \ \
Info on Linux, Graphics, Descent, Laptops, etc. \___________\/
----------------------------------------------------------------------------
Simple End User Linux Leader Mailing list
----------------------------------------------------------------------------