Hi All,
It seems i've sparked a discussion here.. Thanks for fighting the fight for
me Michael :)
Okay.. here is the scenerio..
A sales person takes our 'blackbox' to an office. A requirement is they have
a network and an internet connection. The salesman plugs up the blackbox. If
there is a dhcp server available... hooray! Got the ip address. If there
isn't.. then what?? The device has no keyboard or monitor. Okay.. the
'non-technical' salesman will now have to unplug the box, insert a crossover
cable between his laptop and the device to go to a known ip address to set
the default configuration.. Now you have to make sure every salesman carries
a laptop, which could get very expensive. Better idea.. Use a computer that
is already there attached to their network to push an ip address on to the
box. Once it has this ip address, go to http://the.ip.address.s. and
continue with the configuration.
-Andy
> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED]]On Behalf Of Michael Renzmann
> Sent: Thursday, December 14, 2000 12:00 PM
> To: [EMAIL PROTECTED]
> Subject: Re: arp for setting IP address
>
>
> Hi Wolfgang.
>
> > > Well, *I* have no problems with adding a dhcpd, a bootp-server or
> > > something else to my very own network. But I will have problems if I
> > > force my clients to do so. If someone wants to buy and use the
> > > products I develop, they have to be user friendly. Forcing someone
> > > to extend his
> > Being user friendly could mean that you provide some dhcp/tftp/what-
> > ever-you-need server on the install media for your product. I have
> > bought product that were packaged that way. If you need a tool which
> > the customer will need and might not already have, then provide it.
>
> Being user-friendly for you means that it should be necessary to install
> another server/whatever program, which wants to be configured itself,
> just to pre-configure a device with an IP address, so one is able to
> continue with the "full configuration" afterwards?
>
> > > This is Microsoft style, but Iīm not Microsoft.
> > But you help propagating this silly behaviour by accepting their rules.
>
> Well, this may be depending on the point-of-view. In my opinion it is
> not that harmfull to implement a "backup configuration method" like the
> one described earlier today on the list in addition to those
> standardized methodes you mentioned. Remember, it is (at least not for
> my project, where I could need such methods, too) not the sole
> configuration method, but one of several.
>
> > Don't! Use open standards, and refuse to use proprietary, non-
> > standard solutions (at least when you have a choice - which you have
> > in the case we're discussing).
>
> I do not refuse using open standards. But I refuse forcing the customer
> to use open standards if he has no further need for them. Again: if one
> did not need DHCP or BootP and probably will never need it, why should I
> force him to learn about these things, learn about installation, perform
> the installation of this tool, let him configure it and throw it away
> afterwards just in order to avoid a "proprietary" configuration method,
> which isnīt as proprietary as said?
>
> > I perfectly understand this wish. Hey, I'm the main developer of the
> > PPCBoot (PowerPC firmware) project - I know what you need to set up
> > embedded systems.
>
> I know that fact, we actually talked on that Linux event earlier this
> year in Braunschweig.
>
> > Using a predefined default value; using some type of console
> > interface to manually enter one, using RARP or BOOTP / DHCP.
>
> In the case of Andy there seems to be no console (at least I would not
> call 2 buttons a console to work with for configuring anything...). So
> we are at the point we were before, but I wonīt repeat myself here, as I
> wrote some lines on that above.
>
> > > this task, agree. But forcing people to install them just in order to
> > > assign a 32bit value to a device, or forcing them to tweak the
> > > configuration of a pc for the same purpose if stone-age-behaviour.
> > C'me on, don't tell me what people have to do just to keep their win*
> > systems running.
>
> ?
>
> What should I do if a customer has WinXX running? Laugh at him and cut
> the line? I also think of it (W9x) being a big crap (I know what I talk
> of, as Iīm currently using it on the PC I use to write my mails... but
> only a few days are left for that.. then I will switch again), but I
> have to live with what my customers choose to work with (if they are the
> majority of my customers).
>
> > > So I think this īarp -sī & īpingī thing as proposed by someone on this
> > > list is a fine add-on for the above described cases. If there is no
> > > such tool available by now, I will try to implement it as soon as I
> > > can spend some time on it.
> > So. You call changing a configuration "stone-age-behaviour".
>
> No, read my lines again. Maybe I was not clear enough stating that that
> "arp&ping" thing should not be the only way to assign an IP address, but
> one of several available methods combined within one device.
>
> > Now what do you call re-inventing the wheel again, and badly? I
> > can't help, but at least that's wast of time to me.
>
> Itīs not true reinvention of the wheel (*). At least I donīt know of a
> tool that works like this. Call it "implementation of another method to
> achieve the same thing", with less features than existing methods, but
> also with (assumably or hopefully) smaller footprint than existing
> methods. If one likes to use it, (s)he is welcome to do so. If not, just
> ignore the existence of this tool. Everyone is happy and there is no
> reason to start a flame war on the list. Agree? :)
>
> (*) To keep that example: we do not reinvent the wheel as one knows it.
> We invent a wheel that has the shape of a triangle. It has advantages in
> some cases but might be inconvenient in many others. But no-one gets
> forced to use this wheel. If there is a need for the triangular wheel
> (in a special environment) one can use it, but for driving on a road
> towards sunset it surely is more comfortable to use the old "round" one.
>
> > Wolfgang Denk
>
> Bye, Mike
>
>
> --
> To unsubscribe from this list, send a message to [EMAIL PROTECTED]
> with the command "unsubscribe linux-embedded" in the message body.
> For more information, see <http://waste.org/mail/linux-embedded>.
>
--
To unsubscribe from this list, send a message to [EMAIL PROTECTED]
with the command "unsubscribe linux-embedded" in the message body.
For more information, see <http://waste.org/mail/linux-embedded>.
