Re: Installation issues

--- Philip Blundell <pb_at_nexus.co.uk> wrote:
> On Thu, 2002-07-25 at 20:16, Erik Karlin wrote:
> > I rebooted, and got a syslog about libXi not found...turns out
> > rcS.d/S00bootsplash runs a binary linked to an Xlib. My nvrd
> doesn't
> > get started until rc2.d/S01nvrd, so I made rcS.d/_S00bootsplash and
> > rc2.d/S02bootsplash and rebooted. I don't mind seeing the penguin
> for a
> > few seconds.
>
> I recommend you move the nvrd script to /etc/rcS.d. By the time
> rc2.d
> is executed there is a general expectation that /usr will be
> available.
> You're probably right that bootsplash should run a bit later in the
> order though.
>
> It's a little bit silly that bootsplash is linked against libXi,
> considering that it doesn't actually use X, but the logo itself is
> stored in /usr/share so removing the dependency wouldn't have helped
> here.

Well, just my /usr/X11R6 is on nvrd, but regardless...a triviality

>
> > Another reboot, and I got matchbox. Since I had task-x installed, I
> was
> > left with /etc/rc2.d/S99gpe-dm and /etc/rc2.d/S99x. They do not
> play
> > well together. I hunted around and just moved x to _S99x. Is there
> an
> > update-alternatives or something that could be used...
>
> /etc/rc2.d/S99x is provided by nullxdm, which is supposed to conflict
> with gpe-dm. I'm not sure how you managed to end up with both of
> them
> installed. Or maybe you have an older xserver package which provides
> that file itself.

nullxdm is in unstable, but isn't listed as any dependencies. I just
looked at it's files, and it includes rc[2-5].d/S99nullxdm, and
Provides: x-display-manager. I originally thought you meant that by
conflicting, ipkg would issue an ERROR, and it would "occur" to me that
I'm supposed to remove task-x or something along those lines.

>
> > Next reboot and I just got matchbox again, no appmgr. Turns out
> > libgdk_imlib1 wasn't installed...I assume a missing dependency.
>
> This one is a bit bizarre. Gpe-appmgr shouldn't be depending on
> gdk_imlib any more -- all GPE programs are using libgdk-pixbuf
> instead.
> So I don't quite know whether this dependency is coming from. Mibus
> might have some ideas.
>

no big deal, it's used here and there, possibly "for now".

> > I had dillo already installed, but I don't see it in appmgr...no
> icon?
>
> I have an icon on my machine, but it seems to be missing from the
> package. I'll add it and make a new upload.

Excellent...though not a huge deal. I only brought it up because dillo
was listed as a Depends: for gpe-apps
>
> > Were are all the matchbox utils? I saw gpe-minicalc, but don't see
> it
> > anywhere. What about all the utils that get docked? apm/xkbd?
>
> Click and hold on the dock, you should be offered a menu that lets
> you
> add more stuff to the panel.

I fiddled later, and ended up installing matchbox from unstable(032)
and stumbled on the popup config. Excellent idea. Problem now is that
xstroke doesn't dock and neither does mmenu. I tried adding
/etc/Xsession-gpe.d/65xMenu, but that first puts the mmenu icon at top
left, then it jumps below the dock. Restarting mmenu doesn't help.

>
> > Why is there a gpe-appmgr menu-method, or where is killall?
>
> Killall is provided by busybox.

Yea, the current menu-method for gpe-appmgr works with ps and kill.

>
> > I didn't realize I had to create an account, could not use root.
>
> It's true that you do have to create an additional account the first
> time round, though you can use root for subsequent logins if you
> really
> want to.

That's fine...though I really wish we could hide the pointers in all of
GPE...it's a touchscreen, and the cursor is usually as big as the
embedded xkbd
>
> > If there aren't icons for everything, can there be a default to
> just
> > list the remaining (icon-less) items in /usr/lib/menu, or can we
> run
> > mmenu too?
>
> You can run mmenu as well. Having gpe-appmgr list everything is
> possible, but you end up with a very cluttered view as things stand
> at
> the moment. Gpe-appmgr only has one level of hierarchy, which makes
> things like the "Configuration" section virtually unusable.

Yea, I tried added icons to like xrandr, quite messy...I've no problem
running mmenu if I can get it to dock. I know it docked under
task-x/matchbox/mmenu from v0.5.3.

Now, just stop/start gpe-dm, I have an xstroke "bar" across the entire
screen under my dock. Very wierd.

Anyway, even just a tab for a text listing all the items, perhaps
sorted by section would at least allow access to items w/out icons
without requiring a menu app. Perhaps being able to disable that tab,
and not creating the list.
>
> > Why is ipaq-sleep run from gpe-login.setup with parameters? Why not
> > just use the conf file? according to ipaq-sleep -h, the conf file
> > overrides the parameters...
>
> It's run with parameters so that you get different (more aggressive)
> behaviour when the unit is sitting at the login screen. That
> instance
> of ipaq-sleep is supposed to be killed off as soon as you log in, and
> another one started up using the default parameters or your own
> preferences. Unfortunately, because of a bug in gpe-login.setup,
> this
> isn't happening right now and the initial process is hanging around.

That's a clever idea,and if it's a bug, no big deal.
>
> > What about installing the front end to ipaq-sleep.
>
> Unfortunately the current frontend is written in Python, and we don't
> want to add a python dependency to GPE. Obviously you are free to
> install the frontend yourself if you want. One day I would like to
> see
> the frontend reimplemented in C, at which point we can ship it by
> default.

hmm...sounds like a challenge...
>
> > How about letting ipaq-sleep take an arg for a different conf file,
>
> > then there could be a gpe specific conf file.
>
> Allowing the configuration file to be set on the command line seems
> like
> a reasonable idea, though it's not obvious to me what a gpe-specific
> configuration would really buy you.

Well, I thought the command-line version run from gpe-login.setup was
the one you wanted to run in gpe. It's all clear now.

>
> > I'm quite impressed on how far along things have come. I can't wait
> to
> > start writing some x/gpe code
>
> Thanks for your feedback!
>
> p.
>

A couple more comments.

I really like mbdock, the popup info is very polished looking/feeling.

To-do wouldn't let me add an item until I added a category, and once I
did that, it still only let me add an item once I quit and reran todo.

virt-kbd and virt-kbd setup are a bit confusing, especially yesterday,
I somehow ended up with multiple xkbd's running, and every time I
pressed the dock icon, i'd alternately get the kbd or the setup. Seems
to be alright today though.

With the dock minimized, it's difficult to "click" on it. I reran
xcalibrate a few times, and it's better. It could certainly be a
hardware touchscreen thing.

I also think an auto-hide for the dock is a fairly common feature,
barring any issues above.

minisys update time is really cool, just clicking on xkb and watch the
wave. I guess we'll need some frontend to configure the mini* apps.
Another challenge :)

This page is _very_ helpful in working out gpe:
http://gpe.handhelds.org/projects/GPE-login.shtml

gpe-what sounds like an excellent idea, especially I can't figure out a
reliable way to get the tooltips up. It just doesn't seem to do
anything yet. For now, can it just "balloon" what's in the tooltip?

gpe-login --autolock seems to be running, but leaving my ipaq idle for
quite a while doesn't seem to engage the lock. It did show up when I
resumed this morning.

How do we end a gpe-session? Probably a menu item, huh?

The mapping for the record button to xcalibrate (for gpe-login??)
doesn't seem to work/respond. Either way, it could use a hint on the
login screen. Maybe 3 fields in gpe-login.buttons, a list label, the
xserver sym, and the application. gpe-login could display the label and
the app. or maybe a 4th field for a description.

__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com
Received on Fri Jul 26 2002 - 15:32:45 EDT