[pkg-fso-maint] install.sh vs. d-i (was Re: [patch] Re: pkg-fso-maint Digest, Vol 7, Issue 20)
steffen_moeller at gmx.de
Wed Feb 18 14:58:26 UTC 2009
Michele Renda wrote:
> On 17/02/2009 22:52, Luca Capello wrote:
>> The major problem here is the following:
>> luca at gismo:~$ git clone git://git.debian.org/pkg-fso/files.git
>> Initialized empty Git repository in /home/luca/files/.git/
>> luca at gismo:~$ wc files/install.sh
>> 835 2860 22536 files/install.sh
>> luca at gismo:~$
and it is increasing. My local version is at
$ wc install.sh
1093 3746 30596 install.sh
>> So many lines of shell code are a pain to be maintained, for everyone.
>> And remember we are missing GTA01 support, something which must go into
>> install.sh. We are just duplicating something which is ready and
>> tested, the debian-installer: the more we avoid moving to d-i, the more
>> our (users and developers') time is lost. I will not add more, since we
>> can clearly claim that Debian completely supports the Openmoko phones
>> only when the installation will be done through d-i.
> Ops.... Luca, I think you never worked in a "normal" firm :)
> " I've seen things you people wouldn't believe. Shell script that never
> could see an end.... Java programs less readable than a Korn shell
> script.... etc..."
> When I gave a look on install.sh it seem to be in "heaven" :)
It has functions now :) Since the install process is so very linear, it is truly very
readable. To the trained eye, it seems as readable as concatenated screen dumps of what
one would do manually on the shell. It is close to perfect - for communication among
ourselves - not for the end user. And communication is so very difficult via email, as
well all have previously experienced and probably are experiencing it at the moment. I am
deeply confident that us meeting somewhere (I could not make it for Brussels) would help a
>> Ideally, I would like to delegate all my Debian work to someone, so I
>> could start hacking on the FreeRunner for fun ;-)
>> The pkg-fso team was born and is founded on the do-ocracy principle 
>> (Joachim, please correct me if I am wrong), which means that if you want
>> to take care of install.sh, please go on!
> I think we are focusing too much on the installer. I think installer is
> only a part of the work we have to do here.
> First of all, we have to think how an user must to use Debian, what need
> to be packaged, what need to be write from zero.
> You and Joachim, you have more experiences. I think it will be better if
> you (plural) start to say something like: Michele can you try to package
> this? Steffen can you try to patch this? Someone can please write an
> application to send a sms? Or a new screen keyboard?
agreed! And Luca has by now said often enough that we should look at the Debian installer.
So I will take that one up. I repeat: Steffen will look at using the d-i for the OpenMoko,
and will keep patching install.sh until d-i is doing what its name says. So let us now
concentrate on what to install with either.
> Without this everyone do something, but all the enforces get lost
> because there is no coordination (i.e. I lost 1 months to write an
> application to write an application to make calls on Debian, and then I
> realized that may be only I will use it).
My personal aim is at reliably using GPS. And a good API for the 3D movements I am
interested in. This is why I bought the Neo in the first place, less so for the phone.
But it seems like I am busy with the installers now, so I am relying on you for that :)
> I thing that here we are working without getting money, but only because
> we want to make (in our way) a better world, and we want to be part of a
> I think is important to let feel that the community exists, otherwise is
> very easily to lost the courage to continue.
I have a colleague who is developing applications for his friend's iPhone. It is complete
nonsense, like calculating the time for an egg to cook one you know its weight and length,
or the retrieval of lottery numbers. What shocks me is that Apple managed to come up with
such a fun community. And this is something that Debian should achieve, too. I would not
mind a shop to sell sense or non-sense applications. But for the very moment, we should
truly focus our efforts on getting the core platform fully functional.
Concerning the distribution of work, I kindly suggest/beg Gregor to define working Wifi
environments. He sent an email a few days back that I will prepare a patch for, but I hope
him to finalise it.
More information about the pkg-fso-maint