90% of what I intend to do now and in the future is write web
applications that perform regular business processes (groupware, loan
processing, accounting, etc.), but I also want to be able to write
desktop applications (w/ GUI) using the chosen language. Boring stuff
to a computer scientist, probably, but I'm a former business
professor who became a business drone.
Smalltalk probably has the advantage for these types of apps, yes?
One of the applications will be doing natural language processing on
RSS feeds, Bloomberg terminal feeds, and Usenet postings, etc.
[For everyone who is so interested: It is essentially an application
to combat fraud in the equities markets, and based on govt. funded
research completed while I was at university studying to be a
pointy-haired boss. The primary focus is on fraudulent trade patterns,
but the NLP component would be used to find instances of stock kiting
and shilling that run afoul of investment laws. It would also be used
to find hidden risks in investment syndicates, whether formal or
informal. For example: The failure of the LTCM hedge fund was such a
huge risk to our economic system, in part, because so many of the
investment banks that cleared the LTCM trades were parroting the LTCM
trades. The banks were in awe of the Nobel-prize winning guys that
worked for LTCM..."they're brilliant, so if I do the exact same
thing, I'll look brilliant too!" The risks for the economic system
increased exponentially because of the domino effect of everyone
throwing their weight behind the same trades. One unforeseen
circumstance and everyone gets wiped out...then they had to be bailed
out by the government because it would drag the economy down the
toilet, otherwise. Your tax dollars at work.]
The NLP stuff just needs to be completed overnight, I guess in a
batch-type format...not simultaneously. I'm very interested in
learning more about what Bulent Murtezaoflu was saying about various
approaches and hardware requirements for this type of task.
I'm also interested in creating (relatively simple) expert and
agent-based systems for our company. Lisp probably has the advantage
for these types of apps, yes?
I want to take not only the language features into account, but also
the existing tools, frameworks and code libraries that would help or
hinder my productivity.
I think continuation-based web tools make web development easier, so
I'm trying to compare and contrast the benefits of UnCommon Web and
Seaside. Seaside seems to be a more complete framework, and to have
more lucid documentation. I also know a few companies that are
confident enough with Seaside to use it for production systems now,
while I know very little about UCW. I don't fully understand all the
issues with either, though.
Emacs scares me (there, I said it!) Smalltalk seems to have the more
productive environment...but Lisp seems to have more open-source code
available.
I'm developing and deploying on OS X, with FreeBSD as a possible
deployment option in the distant future. I work with a university
part-time, and I have access to a few racks of XServes...I know almost
nothing about Unix beyond the very basics, though, so the pretty OS X
interfaces provide me some comfort :-)
I'm mainly looking at SBCL or LW for Lisp implementations, or VW as
the Smalltalk implementation. [I prefer to avoid any type of
royalty/runtime scenario, if possible, and (as I said) I'm working on
OS X. These implementations also work with the various frameworks and
libraries I'm interested in so far. I have no problem paying for
commercial tools if they help me get the job done.]
Er...All of my post text made it to the c.l.s list, but some of it
seems to be missing on c.l.l.?
From: Peter Seibel
Subject: Re: Lisp or Smalltalk: Suicide Mission, Part II
Date:
Message-ID: <m3pszzfjbb.fsf@javamonkey.com>
·······@runbox.com writes:
> I'm developing and deploying on OS X, with FreeBSD as a possible
> deployment option in the distant future. I work with a university
> part-time, and I have access to a few racks of XServes...I know
> almost nothing about Unix beyond the very basics, though, so the
> pretty OS X interfaces provide me some comfort :-)
>
> I'm mainly looking at SBCL or LW for Lisp implementations, or VW as
> the Smalltalk implementation. [I prefer to avoid any type of
> royalty/runtime scenario, if possible, and (as I said) I'm working on
> OS X. These implementations also work with the various frameworks and
> libraries I'm interested in so far. I have no problem paying for
> commercial tools if they help me get the job done.]
If you're going to be using OS X you should consider OpenMCL, an
excellent open-source Common Lisp implementation.
-Peter
--
Peter Seibel ·····@javamonkey.com
Lisp is the red pill. -- John Fraser, comp.lang.lisp
Thanks for that suggestion, Peter, but one of the reasons I was looking
at the other implementations to start with is their multi-platform
capability. Down the road, though, I do intend to look at both MCL and
OpenMCL.
From: Rahul Jain
Subject: Re: Lisp or Smalltalk: Suicide Mission, Part II
Date:
Message-ID: <87hdl80z3s.fsf@nyct.net>
········@runbox.com" <·······@runbox.com> writes:
> Thanks for that suggestion, Peter, but one of the reasons I was looking
> at the other implementations to start with is their multi-platform
> capability. Down the road, though, I do intend to look at both MCL and
> OpenMCL.
Why do you think you can't implement your application without an
extremely small portability layer?
--
Rahul Jain
·····@nyct.net
Professional Software Developer, Amateur Quantum Mechanicist
From: Wade Humeniuk
Subject: Re: Lisp or Smalltalk: Suicide Mission, Part II
Date:
Message-ID: <Nz0Id.6474$Qb.6233@edtnps89>
·······@runbox.com wrote:
> [For everyone who is so interested: It is essentially an application
> to combat fraud in the equities markets, and based on govt. funded
> research completed while I was at university studying to be a
> pointy-haired boss. The primary focus is on fraudulent trade patterns,
> but the NLP component would be used to find instances of stock kiting
> and shilling that run afoul of investment laws. It would also be used
> to find hidden risks in investment syndicates, whether formal or
> informal. For example: The failure of the LTCM hedge fund was such a
> huge risk to our economic system, in part, because so many of the
> investment banks that cleared the LTCM trades were parroting the LTCM
> trades. The banks were in awe of the Nobel-prize winning guys that
> worked for LTCM..."they're brilliant, so if I do the exact same
> thing, I'll look brilliant too!" The risks for the economic system
> increased exponentially because of the domino effect of everyone
> throwing their weight behind the same trades. One unforeseen
> circumstance and everyone gets wiped out...then they had to be bailed
> out by the government because it would drag the economy down the
> toilet, otherwise. Your tax dollars at work.]
You mean something like the product from Xanalys (which used to
own LispWorks) and is written in Common Lisp?
http://www.xanalys.com/
http://www.xanalys.com/solutions/financial.html
Wade
It is a different focus area within the same problem domain, I
guess...their product seems to focus on account data,etc. within
individual companies, which is great at finding embezzlement and that
sort of thing, whereas this product would focus more on a meta-market
level where you don't have account information disclosed. It focuses
more on trade clearing, financial instruments/derivative correlations,
volume patterns, etc.
But, yes, I had no doubt that Lisp was suited for this type of problem
- I'm just trying to figure out if Smalltalk is, as well, and if
Smalltalk would offer a lower complexity curve, etc.
From: Wade Humeniuk
Subject: Re: Lisp or Smalltalk: Suicide Mission, Part II
Date:
Message-ID: <hQ9Id.12938$Qb.3653@edtnps89>
·······@runbox.com wrote:
>
> But, yes, I had no doubt that Lisp was suited for this type of problem
> - I'm just trying to figure out if Smalltalk is, as well, and if
> Smalltalk would offer a lower complexity curve, etc.
>
Just to switch the topic back to Domain Specific Languages. Perhaps
you would like to briefly explore what a Lisp-like DSL would like
for your specific language. If you would/could post some very
specific statement from your model, say, a description of fraud
then I could take a stab of transforming that into a Lisp DSL
version. Then you can see how it might be done and what it may
look like.
It the meantime you might visit the site:
http://www.lisperati.com/casting.html
It is a light hearted look at Lisp and maybe why it is
considered to be so good.
Wade
As far as expert systems go there are probably packages you can get that are
compatible with both languages. For Smalltalk you might consider visiting
<http://www.lcdi-tech.com/>. I understand Rich Cooperman has done this kind
of thing multiple times and might be able to help.
·······@runbox.com writes:
> 90% of what I intend to do now and in the future is write web
> applications that perform regular business processes (groupware, loan
> processing, accounting, etc.), but I also want to be able to write
> desktop applications (w/ GUI) using the chosen language. Boring stuff
> to a computer scientist, probably, but I'm a former business
> professor who became a business drone.
>
> Smalltalk probably has the advantage for these types of apps, yes?
I don'f feel so. Both can be used without troubles int those
areas. Some Smalltalks and some Common Lisp implementation even allow
you to run the same stuff on different platforms. I suggest you simply
check out some Smalltalk and some Lisp, read some code and write some
code and *then* might have an impression of what you like or dislike
and can tell your preferences.
Regards
Friedrich
--
Please remove just-for-news- to reply via e-mail.
> 90% of what I intend to do now and in the future is write web
> applications that perform regular business processes (groupware, loan
> processing, accounting, etc.), but I also want to be able to write
> desktop applications (w/ GUI) using the chosen language. Boring stuff
> to a computer scientist, probably, but I'm a former business
> professor who became a business drone.
1) Are you trying to develop a proof-of-concept system, or are you
trying to commercialize a proven concept?
2) Where are the risks? Identify the risks and then choose technology
to address the risks.
When 90% is boring web app stuff, would it make sense to use some
boring commodity tool? WebObjects! PHP!
Why do you also want to write desktop GUI apps? Why wouldn't that
functionality be usable as a web-app? Reduce scope, reduce risk.
How much of that stuff can you get open-source or buy?
Stop this being a suicide mission - understand and manage risk.
3) afaict there's no particular reason the "NLP stuff" needs to be
written in the same language as the 90% web stuff. Seems possible that
you'll want to work on the NLP stuff in an very exploratory way; seems
likely that a lot of the web stuff could be "just another web app".
4) As you're a business professor turned business drone, is learning to
program a good use of your skills?
(At best you'll be a novice programmer, at worst you'll be a really bad
programmer with the authority to enforce really bad design decisions.)
best wishes, Isaac
From: Rahul Jain
Subject: Re: Lisp or Smalltalk: Suicide Mission, Part II
Date:
Message-ID: <87d5vw0z00.fsf@nyct.net>
"Isaac Gouy" <·····@yahoo.com> writes:
> When 90% is boring web app stuff, would it make sense to use some
> boring commodity tool? WebObjects! PHP!
Or use a powerful language and the web app stuff becomes less than 10%,
because you've abstracted HTML nonsensicalities to reasonable
content-focused operators.
--
Rahul Jain
·····@nyct.net
Professional Software Developer, Amateur Quantum Mechanicist
·······@runbox.com writes:
> [For everyone who is so interested: It is essentially an application
> to combat fraud in the equities markets, and based on govt. funded
At the 1998 "40th Anniversary of Lisp" conference, David Lamkins gave
a talk about a validation model written in Lisp for a
telecommunications fraud detection system used by RBOCs.
> I'm also interested in creating (relatively simple) expert and
> agent-based systems for our company. Lisp probably has the advantage
> for these types of apps, yes?
I got several expert systems used books from Amazon at bargain prices.
As for using Lisp, you may check Lisa by David Young:
Lisa - Intelligent Software Agents for Common Lisp
http://lisa.sourceforge.net/
It was influenced by the popular CLIPS and JESS systems. The CVS
version supports uncertain reasoning via certainty factors.
Paolo
--
Lisp Propulsion Laboratory log - http://www.paoloamoroso.it/log
Recommended Common Lisp libraries/tools (see also http://clrfi.alu.org):
- ASDF/ASDF-INSTALL: system building/installation
- CL-PPCRE: regular expressions
- UFFI: Foreign Function Interface
From: Rahul Jain
Subject: Re: Lisp or Smalltalk: Suicide Mission, Part II
Date:
Message-ID: <87mzv00z6y.fsf@nyct.net>
·······@runbox.com writes:
> The failure of the LTCM hedge fund was such a huge risk to our
> economic system, in part, because so many of the investment banks that
> cleared the LTCM trades were parroting the LTCM trades.
Ah, I assumed it was just the sheer amount of money they were managing,
which supported my idea that the investors started pulling their money
out because they didn't realize that they were basically unexposed to
interest-rate risk, but were exposed to liquidity risk. As I understand
it, it was the fact that people got scared and started pulling their
money out that caused them to lose their money. Seems there was more to
this situation that I realized. Yikes...
--
Rahul Jain
·····@nyct.net
Professional Software Developer, Amateur Quantum Mechanicist
Dear Rahul,
> > The failure of the LTCM hedge fund was such a huge risk to our
> > economic system, in part, because so many of the investment banks that
> > cleared the LTCM trades were parroting the LTCM trades.
>
> Ah, I assumed it was just the sheer amount of money they were managing,
> which supported my idea that the investors started pulling their money
> out because they didn't realize that they were basically unexposed to
> interest-rate risk, but were exposed to liquidity risk. As I understand
> it, it was the fact that people got scared and started pulling their
> money out that caused them to lose their money. Seems there was more to
> this situation that I realized. Yikes...
FWIW, my analysis is that LTCM were managing risk by moving risk, while
under the impression they were managing risk by reducing it. A noddy and
big ears example of doing this would be a system for winning at casino by
- bet a point on black
- whenever black loses, double your bet
The formula 2^n+1 = 2^n + 2^n-1 + ... + 1 ensures that every time black wins
you will be one point up, whereupon you begin again. If you have infinite
money to start with, this system does ensure a (pointless) finite addition
to it. For gamblers with finite money, it moves risk from moderate risk
associated with moderately probable events to massive risk associated with
less probable events - in this case, a run of reds sufficient to break the
investor.
With much subtler maths, but arguably not that much more subtle philosophy,
LTCM eliminated risk from the everyday fluctuations of the markets,
delivering steady gains to their investors for some years, at the cost of
accepting massive risk in relation to improbably sequences of events, such
as Russia defaulting on her loans at the same time as a far eastern
collapse, etc. At the point they went under, they had obligations of $3
billion and their system told them that they had simply to assume further
obligations of $100 billion to keep their book level. Similarly, my
hypothetical gambler might tell the casino that if they would only increase
his credit limit by 2^n+1, all would be well. There comes a point at which
reality calls a halt. So while I daresay some at LTCM would argue that
> ... it was the fact that people got scared and started pulling their
> money out that caused them to lose their money ...
I think that would only show their continuing misunderstanding of their true
situation.
Yours faithfully
Niall Ross
From: Rahul Jain
Subject: Re: Lisp or Smalltalk: Suicide Mission, Part II
Date:
Message-ID: <87llai18qi.fsf@nyct.net>
"Niall Ross" <···@bigwig.net> writes:
> FWIW, my analysis is that LTCM were managing risk by moving risk, while
> under the impression they were managing risk by reducing it.
[...]
> Similarly, my hypothetical gambler might tell the casino that if they
> would only increase his credit limit by 2^n+1, all would be well.
> There comes a point at which reality calls a halt. So while I daresay
> some at LTCM would argue that
>
>> ... it was the fact that people got scared and started pulling their
>> money out that caused them to lose their money ...
>
> I think that would only show their continuing misunderstanding of their true
> situation.
I see. I was under the impression that they were mostly doing arbitrage
of almost-fungible securities with very different liquidities. I guess I
was mistaken. :)
Thanks for the explanation.
--
Rahul Jain
·····@nyct.net
Professional Software Developer, Amateur Quantum Mechanicist