|
PureBytes Links
Trading Reference Links
|
Jim, et. al.
There has been considerable disucssion about the advantages of using an
open-source development platform like Linux. Some have even stated that it
"probably represents the only way
to produce complex, reliable software." This implies that those who have
used both Linux and NT consider Linux to be a superior platform and that
alternatives like NT are not as reliable due not only to their archetecture
but also development model. This I find hard to believe for a few reasons.
1. Microsoft generally has access to the best minds in the world. Not only
do coders know they will likely become rich (NT building has among the
highest paid employees at MS) but they know there code and ideas will be
far-reaching. So there's a higher probablity that the best code from the
best minds will work come from Microsoft.
2. I could see where Linux would be typically developed in acadamic
settings, and often a learning platform for the uninitiated and
inexperienced. This really the only thing developers who spend their time on
Linux stand to get from their experience. Since the goal of most software
individuals is to make money, they leverage this experience in a company
like MS.
3. Microsoft has massive internal code reviews so how do these differ and
fall short of the peer reviews for Linux. Coders at Microsoft have to be
directly accountable -- their livelyhood depends on it whereas Linux
developers can just walk away. I could see this as a critical flaw in the
open-source development model -- this translates into less incentive and no
direct accountability for existing problems.
4. Microsoft also produces mission critical code -- their livleyhood depends
on it. If the marketplace is screaming for a feature, Microsoft has a
hsitory of producing software that addresses features requested by the
greatest numbers of people.
But OK, ignoring the above can Linux stand toe to toe with these critical
features of NT (and I'm not an NT expert).
1. app and network stability/recoverability
2. ease of use (GUI)
3. API access (coders)
4. backwards compatiblity (with Windows and DOS)
5. security.
While I don't know a lot about Linux and how the open-source document model
works to handle development hurdles like bug fixes, version control and
synchronization, accountablity it must in someway since it has come this
far. How does Linux address these concerns?
Other concerns are:
1. The software industry desparately needs standardization. Windows gives
us that. It saves programmers time and users time. With ad hoc input from
so many different people how can Linuex continue to standardized new
developments in the future. There has to be a central decision making
commity or no one will agree. Without Stand. We'd back in the DOS daze of
computers when everything was different unfamiliar and you had to spend too
much time learning the interface.
2. Does Linuex have a GUI? Contrary to the belief by some hackers that
computers never really needed a GUI, the GUI has done more to bring
computers into the mainstream than anytying. People think in terms of
pictures and GUI gives us that. The level of art and pictorial
representation is improving with windows. In Linux who will advance the art
and pictorial representation? Different people are going to want to do
things differently. Who will decide? Having a unified company like MS at
the helm helps move us forward.
3. I don't want a private trading platform. I want a public one. I want to
be compatible with the majority. I want to know what the other 90% of the
world out there knows so that when they come to me I can answer there
questions or visa versa. NT5 will do a perfectly good job of giving a safe,
secure netowrkable environment. I want customer support. I want
compatiblity. I want upgrades that I don't have to worry about. With Linux
who decides when a new version is coming out. I want quality standards set
by a company like MS not some ad hoc stuff thrown together by someone and
somebody that may not even care tomorrow about it. If I'm a large company
then I want to use a product that not only is compatible with all other
offices but also I want accountability. Linux isn't made by one company so
there's no accountability. It's buyer beware because you're only solution
if something doesn't work is to hack it yourself!
I know that programmers like it because they can hack the bowels of the OS.
However, this has little utility to the world at large. Below are listed a
few concerns I would have about it. Comments from experienced list members
could benefit all interested parties.
Thanks,
Brian.
-----Original Message-----
From: Jim Osborn [mailto:jimo@xxxxxxxxxx]
Sent: Saturday, October 03, 1998 5:44 PM
To: omega-list@xxxxxxxxxx
Subject: Alternative Development
was: Re: TS5 Press Release
"Brian Massey" <bnm03@xxxxxxx> writes:
>My friend, you don't have a clue what it takes to produce good, commercial
>level software these days. In your transition from hardware guru to trader
>you've lost touch with what it takes to write broad-scale commercial
>software these days.
>
>Andy is 100% correct. Go out an assemble your software team (C++/Windows
>coders only please), spend the $100,000's dollars required to build this
>grade of software in a reasonable amount of time.
The reason Linux has been successful, to the point where it now
represents a credible alternative to Windows NT, is its open-source
development model. With hundreds or thousands of volunteer developers,
who are working mostly for the respect of their peers, rather than to
satisfy a marketing committee, the cost model is different from what
Brian is describing, bugs are unearthed early by teams of eager tester/
developers, and fixed almost immediately. It's a whole different
world from the MicroSoft model, and probably represents the only way
to produce complex, reliable software - massive peer review.
The question for us is, could this open-source development model work
to create a trading platform? I'd like to think so, mostly because
I'd like to use such a platform, and I sort of doubt that I have time
to do it all myself. But I'm skeptical that there's a critical mass
of traders who are also programmers, and are also willing to share
the fruits of their programming in a collaborative effort. Maybe,
but the response here on this list to my troll for a server specification
was a bit underwhelming. :)
Another problem I can imagine is with the data vendors. Even if
a data vendor is willing to disclose details to allow server development,
is that vendor going to choke at the idea of that information being
used in open-source development?
For now, our best prospects seem to be Bob Brickey's TradeLab effort.
It's not open source, but it is by a small team led by a guy who
knows what the words "mission critical" mean, and who has demonstrated
a willingness and ability to do what his customers want. Now if we
could just convince him to port it to Linux, we'd have a real mission-
critical OS under it.
Jim
|