[SoaS] installation fear, was Re: [Marketing] Governance & Trademark in the Wiki

[Douglas McClendon]

Dave Bauer wrote:
> On Tue, Oct 20, 2009 at 5:29 PM, Sebastian Dziallas <sebastian at when.com> wrote:
>> Sean DALY wrote:
>>> Thanks for this Douglas
>> This is all great stuff! :) Thanks for drafting these ideas up...
>>
>>> I believe virtualbox was chosen since free (beer) and multiplatform.
>> Well, it was, from what I recall, since VMware isn't free on Macs and
>> generally not really open source (though they made some steps in that
>> direction). However, the point that has been made on its portability is
>> still valid - I'm not sure how portable it currently is.
>>
>> So here's the thing. I think we might want to provide additional
>> solutions with the Blueberry launch; however, I would prefer not to
>> stretch the SoaS metaphor further than it has already been.
>>
>> https://blueprints.edge.launchpad.net/soas/+spec/joint-soas-release
>>
>> * VirtualBox might be a possibility. I see there's some portable version
>> spreading around, but it seems to be somewhat hack-ish.

Sounds interesting.  Good to know that others are at work on the problem 
and we may soon enough see an un-hackish version by just waiting.

>>
>> * QEMU doesn't require administrative rights and should work directly
>> from USB keys - together with a nice autorun config, this might already
>> fit our needs, though it probably doesn't perform as well as VirtualBox
>> does (mind you, we're already composing virtual appliances at build time).

This sounds good, and like what I'd hoped.  Again going back to being 
less solid than VB.  But since I'm more interested in proof of concept 
than with ease of use, I think I'd prefer that for my own fedora derivative.

>>
> Last I heard QEMU did not have acceleration on Mac OS X so it was
> pretty unusable. I haven't checked into it lately.

While I'm throwing my $0.02 around- note that I brought up windows 
virtualization not as a primary mode of operation, but rather more as a 
weaning stage to help alleviate 'installation fear'.

I.e. for standard windows(/macosx) users, there is no bios/(usb-)boot 
barrier to getting started.  In the event that instead of just being 
slower, it is _a lot slower_, that is all the more motivation for the 
user to read that helpful html they also saw, which can educate them as 
to how easy it is to do that bios/boot fiddling, and become that much 
more in control of the magic box in front of them.

And in the real long term / end-game, the 'real_slow' problems will all 
get fixed.  Maybe there will be an HTML6 <virtualization> tag...

I know some people are eager for 'total world domination / feed all 
starving children' today, but it is important to pick your battles and 
focus on the low hanging fruit that your arms and team are particularly 
good at grabbing.  I.e. I think if you can relatively easily put 
together a sufficient proof of concept, even if it has lots of rough 
edges or is real slow, that is what can help build the momentum and 
attract the wider community needed to hammer out the wider more tedious 
technical problems.

I.e. soas has made a name for itself thus far on what it already is, so 
you don't want to sacrifice that in-hand value for dreams of two birds 
in a bush.  But if you can easily hang onto that bird, but throw a few 
seeds on the ground hoping to attract the other birds...  that is well 
worth doing.

<forgive the parables.... :) >

-dmc


> 
> Do these have the guest additions or whatever for VMWare and
> VirtualBox? For a good experience you want to have clipboard and mouse
> integration with the desktop as well as the video drivers for
> Virtualbox.
> 
> 
>> * Moka5 (www.mokafive.com) - that's an interesting possibility, too, but
>> is only free for Windows, since it requires VMware Fusion on the Mac.
>>
>>> Sebastian, what do you think about these ideas?
>> Well, I think we might want to figure a solution here. Something
>> interesting would probably also be to have it working both on Windows
>> and Macs. If I get a spare minute, I'll think about this and hack a bit
>> into it... anybody up to jump in? ;)
> 
> I can help out with Virtualbox stuff.
>> --Sebastian
>>
>>> thanks
>>>
>>> Sean
>>>
>>>
>>>
>>> On Tue, Oct 20, 2009 at 3:07 AM, Douglas McClendon
>>> <dmc.sugar at filteredperception.org>  wrote:
>>>> Sean DALY wrote:
>>>>
>>>>> You've mentioned how the website could be improved - the "fine print".
>>>>> When you look at the Sugar on a Stick page, what do you think it could
>>>>> express better to guide inexperienced users? The single biggest
>>>>> barrier we face is installation fear - this is how Windows keeps its
>>>>> marketshare (with help from proprietary file formats), and why
>>>>> GNU/Linux desktops have so much difficulty breaking out. Sugar on a
>>>>> Stick sidesteps the problem by not touching the hard disk, but does
>>>>> indeed require system-specific BIOS fiddling.
>>>> In response to this, and DancesWithCars autorun html point, I can see
>>>> possible progress in this direction-
>>>>
>>>> a) autorun html.  Simple to add technically.  I'd opt for pure open source
>>>> but possibly less compatable simple autorun technique, as opposed to using
>>>> the various less-free and often closed source autorun helpers.
>>>>
>>>> b) the content of the html to be autoran- obviously the sky is the limit,
>>>> and something marketing is particularly suited for.  To the extent that
>>>> technical information should be contained, there is the LiveDistro wikipedia
>>>> page, which would be included, as well as a layer above it translated/shrunk
>>>> into a quickstart version targeted at average parents/teachers.
>>>>
>>>> c) other low hanging fruit windows FOSS.  Firefox seems worth it if you've
>>>> got the space.  But more importantly qemu, or whatever the best open source
>>>> windows virtualization solution is (qemu/virtualbox/?). I.e. the webpage
>>>> should include simple instructions for launching that virtualizaiton
>>>> targeted at the CD/USB that contains it.
>>>>
>>>> Now, these are all old ideas I brought up with Fedora years ago, but they
>>>> just aren't that interested, perhaps due to demographics.  I.e. sugarlabs
>>>> demographics would seem to benefit more from these things IMO.
>>>>
>>>> The reason in my own fedora derivative I haven't spent much time on (C) for
>>>> instance, is because I personally just really don't care that much about
>>>> windows.  One thing that scares me is how fragile qemu for win32 sounded.
>>>>   It looks like virtualbox is gpl and available for win32 but I haven't tried
>>>> it.  As such, I think it would be a good idea to do (C), but not really
>>>> advertise it as anything but experimental for at least a year.
>>>>
>>>> Also, since pygtk appears available for windows ala liveusb-creator, perhaps
>>>> the best in the long term would be an autorun program that is just a simple
>>>> pygtk app that can either launch information via a portable firefox install
>>>> pointed at the html on the stick/cd, or launch the cd/stick virtually under
>>>> qemu/virtualbox.  Or enter a chat session with sugarlabs techsupport.  Or
>>>> launch liveusb-creator (in a mode that pulls the data from the stick if that
>>>> isn't yet supported.  I.e. stick replication)
>>>>
>>>> Anyway, thats where I see the lowest hanging fruit for the longterm solution
>>>> to the problem end-users grokking the whole experience upon first
>>>> introduction to the product.
>>>>
>>>> $0.02...
>>>>
>>>> -dmc
>> _______________________________________________
>> SoaS mailing list
>> SoaS at lists.sugarlabs.org
>> http://lists.sugarlabs.org/listinfo/soas
>>
> 
> 
>