Comment 1342 for bug 1

> Date: Fri, 27 Aug 2010 07:33:02 +0000
> From: <email address hidden>
> To: <email address hidden>
> Subject: Re: [Bug 1] Re: Microsoft has a majority market share
>
> On Thu, Aug 26, 2010 at 17:46, Faldegast <email address hidden> wrote:
> > Yes. We should strive to have as much hardware as possible. But still
> > without valuable time and money to support vendors that ignore us.
>
> Full ACK, the vendors that continously ignore the Linux/Ubuntu
> community should be ignored either by the community.
Yeah. People accept that Apple only support the hardware that supports Apple, so there should be no problem doing this with Linux to. I think that we have support for most kinds of devices now.
Lets remember that most of the embedded industry use Linux now. If we start discarding hardware vendors that do not cooperate it will hurt them. It may not hurt motherboard and PC vendors, but it will hurt those that makes the low level hardware. They have their hand both in the embedded and PC market unlike motherboard suppliers that is limited to the PC market.

> > Router corporation does not support Linux? Then we
> > should build our own routers with components that is certified for
> > Linux. There are quite many of us that want such a thing so we should be
> > able to finance it.
>
> I think the financing issue should be solved either (I think you
> touched that issue in your comment either). There is something like a
> portal required where many companies or individuals can share the
> costs of implementing a feature or fixing particular problems/bugs or
> building a driver etc.
There use to be a "virtual stock market" site where people could band together and start projects, that when marketed and sold produced cash for the stock holders. However it was under the dot com boom and died with it.
There are sites where people can share the costs trough bounties. However i think investments is going to bring in more money than donations.

> > Web Services have developed into something really good and fast for remote OOP, once protocols liberated from XML appeared. "Binary XML" like Fast Infoset is currently the best solution for remoting.
>
> Don't know, what you mean with "binary XML", but XML and HTML are
> quite the worst formats ever existed. Why? - Because it is quite
> impossible to build a really efficient parser, nor allow those formats
> continuously append operations. YAML and JSON are already doing a
> better job, but IMHO yet not the optimum. That said, the bottleneck
> mostly is the transfer bandwidth anyway so the more time needed for
> parsing is secondary.
http://en.wikipedia.org/wiki/Fast_Infoset
Fast Infoset and other "Binary XML" formats is basically formats that are on a API-level compatible with XML parsers but stor data in a binary data structure. As far as I can determine Fast Infoset is the fastest thing around currently. Its a little perverse that they have XML in their name because they dont really have anything to do with XML... but as XML is the hype it may be needed to fool project managers that they are not missing out on the XML hype...

> > I think its a bit amusing that Web Services that was supposed to be XML
> > and easy to read is developing into a fast binary protocol like RPC. In
> > the end speed does matter.
>
> E.g. SOAP is one of the worst things I have ever seen. I hate it.
I agree there. I dont even know how it really worked. I read trough the overview when the hype started and stopped there. The only problem it solves is high portability, but it creates more problem then that...

> > I think we really need the input from some kernel hacker here.
>
> And here probably is not the best location to continue this
> discussion...
Perhaps. But where is the best location?

> > Also designing something to work on all platforms including Windows will
> > help bridge the #1-bug as it may lure Windows developers away from COM
> > and other Windows-specific technology.
>
> Indeed, one of the most important things at this point in history is,
> to offer solutions that work everywhere so Windows (or Mac) users can
> try and when they see, that they can work with the same applications
> even if they would switch to Linux/*buntu they would be more likely to
> give it a try and enjoy.

Also in enables Linux-Windows interoperability. Applications running on Windows clients/servers can transparantly call Linux clients/servers and vice verse. For larger multi-server applications that allows a slow migration to Linux. Now components can be implemented on Linux while old ones stay on Windows.

> > Having high market shares in five years is not always as
> > important as providing a good report this quarter. They may not even
> > work at Microsoft in five years, but need their CV to look good.
>
> The slow and steady success outweighs the short-term success...

Not in the mind of the individual suits that desire to put that short-term success on their CV... We tend to think of Microsoft as a person when it in fact are a group of individuals. Some say that an enemy is never stronger then its weakest link. If we limit "the enemy" to only include those that have the authority to make this kind of decisions this is very true.

So while the slow and steady success out-weights the short-term success for Microsoft as a whole, individual executives will act in their own best interest, not in Microsoft's.

> > Migration to a new version of Ubuntu is a lot less painless if you can
> > run it side by side with the old version. Live migration is also cool
> > but of limited value on the desktop.
>
> I never run side-by-side - simply just because: Why waste HD space -
> even if it is just 4 GB? No matter how big a HD is, it tends to be
> full anyway (just like basements). ;-)
There is two kinds of disks. New disks and full disks. :)
Well the idea is that once you know the new version works you just delete the old one. You would never... well... almowt never have to reinstall because an upgrade fail.
Also a lot of work is put into the Ubuntu upgrader and still its never guaranteed to work. By just creating a new OS image with the new version upgrading becomes simple. And as both versions initially run side by side you will not end up with a failed update that does not work.

> I think, we are going to much in detail to be discussed here, but
> anyway, good that those things have been mentioned...
Yeah, probably. :)
> --
> Martin Wildam
>
> http://www.google.com/profiles/mwildam