Re: can't install wwwoffle

2008-11-06 Thread Alastair Johnson

[EMAIL PROTECTED] wrote:

Trying to install wwwoffle, after installing cron
I still needed to add "debian-gta02" to /etc/hosts:
127.0.0.1 localhost debian-gta02

However the install still gets stuck generating
certificates/root/root-key.pem, so one kills it, leaving a zero byte
file. A subsequent apt-get install gives:

The WWWOFFLE root CA files don't exist. Please stand by while these are
generated (this may take very long, depending on your system).
wwwoffled[12002] Important: WWWOFFLE Demon Version 2.9d (with ipv6,with 
zlib,with gnutls) started.
wwwoffled[12002] Warning: Could not import private key [Base64 unexpected 
header error.].
wwwoffled[12002] Fatal: The WWWOFFLE root CA private key file 
'certificates/root/root-key.pem' cannot be loaded; delete problem file and 
start WWWOFFLE again to recreate it.
invoke-rc.d: initscript wwwoffle, action "start" failed.
dpkg: error processing wwwoffle (--configure):


Are you sure it's getting stuck? Without a good source of randomness key 
generation can take a _very_ long time.


___
Smartphones-userland mailing list
Smartphones-userland@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/smartphones-userland


Re: [Debian] install.sh: adapt to fso-config-* packages

2008-11-07 Thread Alastair Johnson

Luca Capello wrote:

Hi Joachim!

On Thu, 06 Nov 2008 21:02:08 +0100, Joachim Breitner wrote:

Am Donnerstag, den 06.11.2008, 19:57 +0100 schrieb Luca Capello:

   - /etc/default/rcS

 By default we enable fsck at boot, is this a good thing?

[...]

The major problem I found with this setting is that when the FR runs out
of power, as soon as you plug in the cable (being it the USB or the wall
charger one) Debian starts and then fsck fires up, which results in the
FR running again out of power.  The only solution I am aware of is to
switch on the distribution installed on the internal flash memory.

It runs out of power despite being on the cabel? Hmm, that’s bad.


AFAIK this is a known and expected behavior [1], because the FR requires
more power than the one provided by the wall charger when the battery is
completely empty.


Not in the situation described. To be doing fsck the kernel must be up, 
so it should have set 500mA charge mode for USB, or 1000mA mode for the 
wall charger. This is enough to power the FR _and_ charge the battery.



But what else should we do? Having the filesystem slowly corrupt over
the time is no a good solution for a Debian system either, is it?


I am for all in favor of fsck at boot, but not everytime.  Enrico's
proposal at [2] could be a step in this direction, I will post my
comments there.

Thx, bye,
Gismo / Luca

Footnotes: 
[1] http://wiki.openmoko.org/wiki/Forcing_fast_charge_mode

[2] 
http://lists.linuxtogo.org/pipermail/smartphones-userland/2008-November/000374.html


___
Smartphones-userland mailing list
Smartphones-userland@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/smartphones-userland


Re: [Debian] install.sh: adapt to fso-config-* packages

2008-11-11 Thread Alastair Johnson

Enrico Zini wrote:

On Fri, Nov 07, 2008 at 10:48:11PM -0500, Stefan Monnier wrote:


I haven't seen actual concrete (i.e. not just hearsay or theoretical)
reasons to prefer ext2 to ext3 on flash.


I have friends that work with embedded hardware and tested CF and NAND devices
with ext3, and their experience is that on flash devices which don't implement
leveling properly, after one day the system became unbootable because of I/O
errors in the journal.


The distinction here is that pure NAND generally will not have write 
levelling, while CF and SD almost certainly will. I have been using ext3 
on CF for several years now with no such problems; they have been more 
reliable than HDDs so far. Perhaps I have just been exceptionally lucky.



Yes, this qualifies as hearsay for everyone except me who got the news directly
from the source, so feel free to ignore me.  However, I trust my source enough
to object to having ext3 as the default or the suggested file system in Debian
setups.


Fair enough. It's direct experience for me, but hearsay for everyone 
else. I'll keep using ext3 though.


___
Smartphones-userland mailing list
Smartphones-userland@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/smartphones-userland


Re: Building FSO

2009-02-06 Thread Alastair Johnson

Arigead wrote:

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1


Hello all,
I was just trying to build the FSO from openembedded as per the
instructions on the wiki. If you're trying to do that on an eeePC it
tends to take some time! I only have to build it all once and then all I
have to build changes. Anyhow the build has failed at connman, a few
days in ;-)

I can't see what depends on connman and why I'm even building it.


connman is (meant to be) a lightweight equivalent to NetworkManager, and 
is going to be used to handle networking.



I'm not entirely confident here that what I'm building in openembedded
will match the MS5 build that I have in my phone at present. As far as I
know the official FSO is built from openembedded upstream but I don't
think that this can be the case.

I've said before that the openmoko should have there own small
repository for bitbake recipes that can overload the official upstream
openembedded recipes. Those should be kept in a directory at the same
level as openembedded and BBPATH used to set priority. Then
om_openembedded can overload any upstream recipes as it sees fit and
produce recipes that build the latest kernel etc. I imagine my build of
fso-image would not have had the andy tracking kernel but an old one.

I'm thinking of a structure like the gumstix structure [1]

Anyhow those are just suggestions, in the mean time can anybody tell me
how I can build fso-image as per the wiki instructions?


wget http://downloads.freesmartphone.org/Makefile
make setup

You could now enable parallel builds in common/conf/local.conf but this 
can cause problems with some packages, and probably isn't a good idea on 
the Eee anyway.


make fso-gta02-milestone5-image

I had to rerun this a couple of times as a couple of packages failed due 
to parallel build issues and lack of memory. When you rerun it only 
builds the bits that haven't successfully completed yet.



Thanks for any help

[1]
http://www.gumstix.net/Software/view/Build-system-overview/Directory-layout/111.html
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkmL9bEACgkQXlbjSJ5n4BD3cgCgo+v0V7LJ6LV2v57ubo2Q9swH
rUsAoJ0CpPvABz1WF6aJRWGCp+lUKE5j
=hKMg
-END PGP SIGNATURE-

___
Smartphones-userland mailing list
Smartphones-userland@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/smartphones-userland



--
SolutionTrax Limited, Registered in England & Wales, Trading as TrueBox
Registered office: Silbury Court, 420 Silbury Boulevard, Milton Keynes, 
MK9 2AF, UK

Company Number: 4494022
VAT Registration Number: 825 1952 26
Telephone: +44 (0)845 508 3397

___
Smartphones-userland mailing list
Smartphones-userland@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/smartphones-userland


Re: cacao on fso 5.1 does not work

2009-03-13 Thread Alastair Johnson

Davide wrote:

r...@om-gta02:~# java -version
Absurd stack bottom value
Aborted


Confirmed with cacao and deps from the fso-testing repository. java 
-fullversion works correctly, but everything else I've tried, including 
trying to run a HelloWorld app that works elsewhere, gives the above 
message.


I haven't tried with the binaries from evolvis yet, but it's probably 
better reported there than here.


___
Smartphones-userland mailing list
Smartphones-userland@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/smartphones-userland


Re: [Debian] How do I change the ringtone again?

2009-03-24 Thread Alastair Johnson

Enrico Zini wrote:

So I tried this (from ssh):
  # /etc/init.d/nodm stop
  # /etc/init.d/fso-frameworkd stop
  # /etc/init.d/fso-frameworkd start
  # /etc/init.d/nodm start
I got asked for the PIN again and the phone started ringing (finally
with the right ringtone) while at the other end it had started to say
"You have dialed an incorrect number, please check the number and
redial". It seems that the phone keeps registering and unregistering
itself, greatly confusing the UK mobile phone network.


Are you using O2? Reregistration caused by bug #1024 seems to confuse 
their network sometimes, but t-mobile and orange don't seem to mind. If 
it is #1024 you should be able to work around it by adding a line to 
frameworkd.conf in the [ogsmd] section:

ti_calypso_deep_sleep = never

___
Smartphones-userland mailing list
Smartphones-userland@linuxtogo.org
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/smartphones-userland