Re: Proposals and information

[Ken Yap <ken@nlc.net.au>]

Comments on Telford's post:

First a general comment. What Kylie did was attach two separately
conceived proposals together. As noted above the second proposal,
that was developed with a particular organisation in mind, not for
computerbank. This accounts for any apparent discrepancies and they
should be read in this light. Personally I think Kylie's proposal is
the one that should go to high level committees. It's only when we come
to implementation that we can debate the finer points mentioned in the
second.

>The only choice is to accept everything and test it yourself, with no idea
>of where it's comming from there is no way to guarantee quality of supply.
>Most dealers of NEW equipment can't guarantee quality of supply!

I believe this is the correct. Most givers will not be able to say much
about the quality ("I heard you take old computers and my boss told me
to get rid of these old boxes").

>386 machines. Whatever way people want to go on the 286 machines, the organis
>ation
>aims should not be wishy-washy about it.

Details like 386 should be left out of the high-level aims because
some givers won't even know what is a 286 and what is a 386. Which is
why skip(s) behind the workshop area and an arrangement to collect at
intervals are important.

286s can be turned into diskless terminals in a network setting. See my
page at

	www.geocities.com/SiliconValley/Lab/9247

>One of the advantages of open-source software is that having visible source c
>ode
>is an aid to learning. I realise this is understood but it's not made clear
>by this document.

I doubt if most of the users of computerbank want to be coders,
but it's there if they want it. Rather the superiority of OSS comes
from the support arrangements. Say we adopt some OSS package to help a
disabled person.  In effect we are expanding the tester base for that
code. And if the maintainer decides to pack it in, well somebody else
can take over the torch.

>model
>which can be adapted to the Computer Bank purpose? I was thinking something a
>long
>similar lines to TV and video rental with rental rates being determined on a 
>case
>by case basis depending on ability to pay. This gives it a bit of a commercia

As you may know, TAD charges a nominal support fee per year for lending
a machine to a person.

>> * Promote the use of open-source software, thereby using the technical
>>   skills of locally available expertise while minimising total and
>>   ongoing end-user costs. (S)
>
>Probably (L). Computer Bank should be seeing GNU/Linux, etc as a
>means to an end not an end in itself, even though individuals may see
>it the other way around. Maybe this goal shouldn't even be here.

Ah, OSS > GNU/Linux.

>rubber carpeting? Static electricity is probably going to be the
>main problem. If electrocution is a worry then better than rubber
>carpet are RCD (earth leakage) switches on each circuit and a big
>red OFF button in easy reach on each bench to kill the power at that
>bench. Also some safety procedures that are written up nicely around
>the place: e.g. spend the last 15 minutes of each shift tidying up
>the benches, keep walkways clear at all times, equipment found to
>be faulty should immediately be labeled -- there are plenty of these
>kind of documents floating around in industry (though I'd guess that
>most of the computer assemblers in Sydney ignore them) maybe also
>bother the TAFE workshops see if it's OK to copy whatever they
>are doing.

A lot of fuss is made about static and the ability to destroy electronic
components. In general the components are at risk when they are out
of boards. On boards they are quite hardy. I have mistreated boards in
many ways.  I have burnt many chips, but I have not had any failures I
can put down to static. The exception may be memory sticks. A grounded
wrist strap will do fine.

>>   Software would be alternatively installed from CDROM(s), 
>>   which would contain a compilation of open-source and free software
>>   required to (re)create a end-user setup.
>>   We would provide a copy of this CDROM with every shipment.
>
>Off the top of my head I'd guess that (ignoring 286 machines)
>about half of the gear that we scavenge would have CDROM drives.
>Providing a copy with every shipment might not be always useful,
>maybe selling them at a low cost to whoever wants one.

Here is where the different origin of the documents should be taken into
account. The reverse computer concept envisaged shipping entire networks
of computers and having a local person maintain that installation.

CD-Rs are incredibly cheap these days ($21 for 10, jewel box and all). I
have a writer.

>> * A Repair area.
>>   If there is sufficient skill available and a suitably qualified 
>>   instructor, repairs can be done for those items that are repairable.
>>   This will necessiate the need for a small inventory of components and
>>   the availablity of "petty cash" or an account with component stores.
>>   Such products as monitors, keyborads, modems, cables and printers
>>   would be repairable most of the time. Of course, first component
>>   requirements would be forefilled by "ratting" another dud unit.
>
>Repair of monitors is OK but it's tricky work, you don't want to
>do it alone and training people to do it won't be easy.
>
>Repair of swich-mode power supplies should be just replacing the
>internal fuse, if the fuse blows next time you power it up then throw
>the whole lot away. If the supply looks functional, check insulation quality
>on the switch and cables just in case. There isn't a lot more that
>you can do with them :-( Is anyone aware of people who do more than this?
>
>Repair of hard drives is possible but again, highly specialised and
>usually not worth the effort.
>
>Some repair on keyboards actually is quite easy. Usually they just need
>to be taken apart, cleaned, dusted lightly with powdered graphite and
>put back together. It takes about 20 minutes if you don't lose any
>of the millions of bits that can go everywhere. Again, barely break-even
>in most cases. I only do it for keyboards that as specially valuable
>such as those with a long space bar and without Microsoft-logo keys.

I should have read the reverse computer document more carefully
before it went out. I do not now believe those components are worth
repairing. Printers maybe worth the effort if they are laser. Otherwise
it takes too much skill.  We should just toss the ones that malfunction.
And chances are, we can satisfy needs by mixing and matchng.

>All up I'd say that anything broken should be tossed out. It's wasteful
>but time is precious. A bigger problem is gear that probably does work
>but documentation on jumper settings or configuration is not available.
>Linux drivers are available for most common hardware but usually this
>involves knowing and setting IRQ and IO lines. Sometimes it's worth
>trying possibilities in sequence till something works, often it's not.

This is actually less of a problem than it may seem. Firstly most common
boards are already at their default settings so we just document it with
a sticker and leave it as is. We will have enough samples to pick boards
to suit the box. NICs are the most troublesome, but I have learnt a lot
in this arena, see my page at:

	www.slug.org.au/NIC

>We are going to have drives with unknown C/H/S settings that won't respond
>to IDE enquiry because they are too old, BIOS passwords that someone has
>set long ago and forgotten, motherboards that require alternative jumper
>settings when their memory is changed, all sorts of nasty stuff.

Most drives that don't respond to inquiry are dead of old age now. I
wouldn't touch anything smaller than 80 MB or so. BIOS passwords can be
fixed by disconnecting the battery temporarily (and weak batteries are
likely to be a problem anyway). The settings are usually correct for
what's on them and can be left as is. It's less of a problem than may
seem. I have rescued many MBs and hardly ever needed to touch any jumpers
(except the Mono/VGA) one.

>> Our standard offering is based on a central Linux server providing
>> all sorts of network services. Around these we cluster machines each
>> belonging to a category of capability. The category also determines
>> more or less what sort of hardware we put on it.
>
>Are whole networks really the target? I'd guess that more stand-alone
>machines with a modem for homes would be better. To be honest I can't
>say that I know what the demand really is. Some study into this area
>would be pretty useful.

Again, factor in the different goals of the original document.

>> Need GUI admin programs for workers to load up workstations.
>
>As does all Linux, it's comming slowly though...

GUI != X. And network install means a lot can be done from a central
server.

In summary, I think the computerbank document should be tweaked to
remove distracting details. I would edit it thus:

Remove the paragraph about what we will or will not accept and other
stipulations on what we take. Those are details and will have to be
enforced at the time of donation if at all, or we just allocate a
larger skip.  In particular remove the 'radiation vibes' thing, it
reads unscientific.

The advertising thing may be in conflict with public sector support. Also
the Internet Web business is dubious. I'd drop those.

I believe equipment is actually the least of our hurdles. You will not
believe some the useful stuff I find in tips. The main hurdles will be
enough people to help. Witness TAD with 500 computers and not enough
helpers. Which is why reverse computer envisaged assembly line and
automated software installation techniques to make effective use of
volunteers' time. Another kind of volunteer needed are people to teach
users.  That is why computerbank should enlist other organisations and
individuals to address the user handholding aspect. We tend to assume, as
computer literate people, that computers are easy with one hand holding
a guide and another on they keyboard. A teacher friend of mine can tell
you different.

And while we are pursuing high-level support, we can start in a practical
way. My friend knows of some needy users and I know of some potential
donors. Now we are looking for a place to refurbish equipment. Maybe
his storeroom.

Another fruitful place to try might be schools. I'm installing a Linux
web proxy in a public school next week. In time I'll show them that other
pieces of donated equipment could be turned to use on their network in
various ways.

	Cheers, Ken