Gartner Group vs Smalltalk
Taylor Corey
Smalltalk is a niche market and some people/companies do not want to play in
that arena. I said before the number of VAST licenses totals 80K, I'll
retract that because I think the real number is more like 8K.
Some major cities like Atlanta have no ST going on at all to speak of and
the Atlanta ST User's Group web site is no longer. One way to see how big
ST is would be to get one of the languge reports from the Gartner group
which looks at license revenue for various languages. These reports cost
several thousand dollars. So I have to resort to public sources of
information for my analysis. Before I do that, one should note that the
Gartner Group says that Java is a lot like ST and even they predict that IBM
will eventually migrate everything they can to Java. Here is my perspective
on the ST market:
OBJS' 1995 license revenue was $32 million (http://www.sec.gov/ search the
edgar database for objectshare). At $5K per license (which was the price of
VW back then, I think), that comes to 6000 dev licenses which is high since
some of the license revenue came from runtime fees. If st is 10% of all OO
dev activities (from a 12/97 Gartner Group Report, go to
http://www.gartner.com/ and do a search on "smalltalk") then the total
amount spent on OO would $320 million. If the total IT budget for the US is
$200 billion then the percent spent on OO activities would be neglible.
So the next question is where does the $200 billion number come from? In an
article by ComputerWorld at:
http://www.computerworld.com/home/print.nsf/all/000103D7C2
US costs for fixing y2k is $115 billion and in the same article, they report
that companies are spending anywhere from 5% to 40% of their IT budget on
y2k. I don't think it would be unreasonable to say the current total US IT
budget is around $300 billion. Lets say 1% of that $300 billion is spent on
OO activities (JAVA, C++, ST, etc.) or $3 billion.
A 6/99 report from the Gartner Group says that ST will make up less than 5%
of all OO activities or $150 million (5% of $3 billlion). Lets say IBM gets
4% of that $150 million with VAST, that would be $60 million. Still a large
market and at $5K per dev license, that's 12,000 licenses.
Looking at the Java side of things, 'Java Developer's Journal'
(http://www2.sys-con.com/recordcirculation.cfm) reports a circulation of
70K. I think most people would be hard pressed to find a smalltalk magazine
let alone one with a circulation half that amount.
--tc
Bijan Parsia
[snip]
> Some major cities like Atlanta have no ST going on at all to speak of and
> the Atlanta ST User's Group web site is no longer.
[snip]
I'm not sure commerically speaking (though I do believe I've seen
Atlanta based Smalltalk offerings), but GA Tech is the second major home
of Squeak development (IMHO). So, I'd like to see a little more evidence
before accepting this claim.
After all, what one cares to speak of surely varies with the speaker.
Cheers,
Bijan Parsia.
Taylor Corey
Medical. They have since abandoned all st development. The company that
bought them, McKesson, has some st development in San Francisco (I believe).
There is Melita in Norcross, Mediphus near the Galleria. The bottom line is
it is not very much considering the size of Atlanta and not even enough to
keep the user's group going.
--tc
Bijan Parsia <bpa...@email.unc.edu> wrote in message
news:1e4vngn.1s3tmhp1xlsbsrN%bpa...@email.unc.edu...
DOUGLAS A SWARTZ
Taylor Corey <tco...@mindspring.com> wrote in message
news:86fsvr$akk$1...@nntp4.atl.mindspring.net...
> I have the same question. How does one become an effective advocate for
ST?
> Smalltalk is a niche market and some people/companies do not want to play
in
> that arena. I said before the number of VAST licenses totals 80K, I'll
> retract that because I think the real number is more like 8K.
>
includes hardware, all software developer and consulting salaries, system
administration and other costs. I think you're trying to compare apples and
oranges.
Doug Swartz
dasw...@prodigy.net
G. S. Link
VISUAL development platform so that non-computer people can write simple
programs such as GUI interfaces to databases. These people know no
Smalltalk and do not participate. I suspect that is where the majority
of Vast usage is. These people do not use VaJava because the visual
builder won't support what they want to do.
Taylor Corey
but if software licenses are a smaller part of the IT budget as you say,
then that would relegate Smalltalk to a much smaller niche market. I doubt
that VAST licenses are below OBJS' peak of 6K - 8K but they wouldn't be much
more either.
--tc
"DOUGLAS A SWARTZ" <DASW...@prodigy.net> wrote in message
news:86hkbi$sk0$1...@news.software.ibm.com...
Frank Sergeant
> then that would relegate Smalltalk to a much smaller niche market. I doubt
> that VAST licenses are below OBJS' peak of 6K - 8K but they wouldn't be much
> more either.
But, you are talking about the number of licenses sold per year, right?
So, you need to mulitply that by the number of years to get an idea of
how much copies have been licensed. (Then, apply some sort of ratio
to see how many of those licenses are used rather than sitting on a
shelf.)
Is There a Doctor in the House?
(Of philosophy, that is) While we are on this general subject,
I wonder if Bijan or others have looked into the Cincom approach
to licensing VW commercially. I had a nice talk on the phone with
Cincom about two weeks ago and had the different options explained
to me. The only one that would seem to fit me would be the VAR
package, would would involve paying royalties. It doesn't sound completely
unreasonable but, at the same time, I have a reluctance to agree to
paying a royalty. (And, it wasn't quite clear from the phone
conversation exactly what the amount of the royalty would be or
exactly what the terms of the contract would be that I would need
to sign. Eventually, perhaps I might try to get answers to those
two questions.) So, what are your thoughts on this? Should I
jump at a chance to make Cincom my "partner" or not? Would any
philosopher in the audience offer a strategy for thinking about
this problem?
-- Frank
frank.s...@pobox.com
tco...@mindspring.com
Frank Sergeant <frank.s...@redneck.net> wrote:
>But, you are talking about the number of licenses sold per year,
>right? So, you need to mulitply that by the number of years to get an
>idea of how much copies have been licensed. (Then, apply some sort of
>ratio to see how many of those licenses are used rather than sitting on
>a shelf.)
OBJS' license revenue dropped way off after that peak and some of that
revenue came from runtime fees and not a dev license. I would still
maintain a number between 6K and 10K active dev licenses. The 8K number
was for their one year peak.
--tc
Sent via Deja.com http://www.deja.com/
Before you buy.
Taylor Corey
License Revenue for OBJS (in thousands), '95 to '97 is from the 10K, '92 to
'94 is just a guess at the total combined revenue for ParcPlace and Digitalk
because no 10-K is available.
1997: $20,148 or 4,000 dev licenses
1996: $31,667 or 6,000 dev licences
1995: $32,521 or 6,000 dev licenses
1994: $30,000 or 6,000 dev licenses
1993: $30,000 or 6,000 dev licenses
1992: $30,000 or 6,000 dev licenses
Total: 34,000 dev licenses
That seems like a lot plus some of the revenue comes from runtime fees. If
someone told me there were 10K or even 20K active licenses, I might believe
it but when I was at KSC, I don't remember hundreds (like 700, 800, or 900)
or thousands of companies using smalltalk. Plus some companies did 'one
off' test projects.
--tc
"Frank Sergeant" <frank.s...@redneck.net> wrote in message
news:87d7qrs...@den.home.net...
> "Taylor Corey" <tco...@mindspring.com> writes:
>
> > then that would relegate Smalltalk to a much smaller niche market. I
doubt
> > that VAST licenses are below OBJS' peak of 6K - 8K but they wouldn't be
much
> > more either.
>
> But, you are talking about the number of licenses sold per year, right?
> So, you need to mulitply that by the number of years to get an idea of
> how much copies have been licensed. (Then, apply some sort of ratio
> to see how many of those licenses are used rather than sitting on a
> shelf.)
>
>
Bijan Parsia
[snip]
>
> Is There a Doctor in the House?
>
> (Of philosophy, that is)
Er...Not yet! But if a mere M.A. *in* philosophy will do, I'm at your
service :)
> While we are on this general subject,
> I wonder if Bijan or others have looked into the Cincom approach
> to licensing VW commercially.
Er...only what I've read in the group, and gleaned from the website (I
am wondering what a "VisualWorks Ambassador" is). I've been *planning*
to, of course ;) But I'm sorta waiting to see what happens with the NC
licence.
> I had a nice talk on the phone with
> Cincom about two weeks ago and had the different options explained
> to me.
Hey! Explain 'em to us! :)
> The only one that would seem to fit me would be the VAR
> package, would would involve paying royalties. It doesn't sound completely
> unreasonable but, at the same time, I have a reluctance to agree to
> paying a royalty.
As do many. But if you're accounting can handle it without much
difficulty I wouldn't think this should be a dealkiller.
(But *note* Cincom folks! It *is* a barrier. Even people *willing* to go
along with it feel *reluctance*! You don't want that! Make it easy. If
you're going to have royalties, make 'em clear as day! Look at the
licencing of Macintosh Common Lisp (MCL) http://www.digitool.com. They
seem quite reasonable, and unconfusing. Don't call them royalties, call
them "distribution" rights and limit them to a dollar amount, number,
geographic region, organization, or the like.)
(Back to our regularly scheduled advice :))
> (And, it wasn't quite clear from the phone
> conversation exactly what the amount of the royalty would be or
> exactly what the terms of the contract would be that I would need
> to sign. Eventually, perhaps I might try to get answers to those
> two questions.)
(Ack! Cincomers! This shouldn't be! One phone call should generate more
followup documentation than you can shake a still at!. He has to make a
*followup* call? Sheesh! Do you *want* to sell this product!
We return, yet again, to our regularly scheduled advice.)
> So, what are your thoughts on this? Should I
> jump at a chance to make Cincom my "partner" or not? Would any
> philosopher in the audience offer a strategy for thinking about
> this problem?
Very very much depends on what you're doing and why, but my personal
basic thought is: If your task just *screams* for VisualWorks, and the
pricing structure doesn't kill the deal, then go for it. Dithering is
likely to be as wasteful as plunging, and it's a lot more painful.
If it's no big deal either way, then try to negotiate. If Cincom is
going to take the approach they're taking then you should expect
personal service and accomedation. For example, if they are going to
demand royalties on a product you're distributing, they'd should be
ready to help you handle tech support problems (that arise from Cincom
stuff, obviously).
(Personally, I think Cincom should divorce licence fees and tech
support, and be liberal with the licences. You needn't GPL something to
make your money of tech support.
MCL is a good example, with their "News-stand" & "champion" licences.)
But, in general, if they are going to be your partner, hold them to it.
Think about what you'd expect and want from a partner, and see if
they'll do that.
Cheers,
Bijan Parsia.
Taylor Corey
responsible for a large project your company wants done. The Gartner group
tells you that Java was heavily influenced by Smalltalk implying they are
equivalent and the popularity of Java means you will not have trouble
getting people. Here are some numbers:
Conferences
=========
1. Smalltalk,
- 225 paid attendees at st solutions '98, http://www.stic.org/STIC/SS99.HTM
- OOPSLA 98, STIC had a reception with 500 people,
http://www.stic.org/Adventure/B015.htm
- 1000 people attended the PPD International User's Conference according to
the 9/95 ST Report.
2. Java
Sun's 1999 Java one conference had 20K people,
http://java.sun.com/features/1999/06/photoalbum1.html
Magazines
========
1. Smalltalk,
- An article in JOOP.
- St Report went out of circulation. In its best year, I don't think it
came close to 70K subscribers but someone correct me if that is wrong.
2. Java,
- 'Java Developer's Journal' has a subscriber base of 70K,
http://www2.sys-con.com/recordcirculation.cfm
Most people like to go with the flow, so which environment are you going to
go for?
--tc
Thierry Thelliez <thi...@acm.org> wrote in message
news:3888BA21...@acm.org...
> Dear ST colleagues,
>
> Not a good week for us here. My management decided to get the advise of
> the Gartner Group
> about IM tools. Of course Smalltalk (and GemStone/S) are not their
> prefered tools.
> GemStone/S was described as 'questionable' and no new projects should
> start with Smalltalk.
>
> Any idea on how to counter attack that ?
>
>
> Thierry
>
>
Donald Malcolm MacQueen
Taylor Corey wrote in message
<86ivre$r5t$1...@nntp6.atl.mindspring.net>...
RIght. They used to say in the hardware business that,
"No one ever got fired for buying IBM". Java is now in
that position in IT. And, as Dan Ingalls pointed out,
"Java is a step up for the C++ crowd".
--
Donald
----------------------------------------
Donald M. MacQueen
Registered Smalltalk Bigot
----------------------------------------
I've tried the rest, now I use the best -- Smalltalk!
----------------------------------------
Dan Antion
fairly rapid DOS based development environment (Clipper), spent slightly over
6,000 hours developing a rating system for our liability product. In 1999, we
spent slightly over 1,200 hours completely rewriting that system in Smalltalk.
The base functions of the system did not change, although many features were
added to the Smalltalk version.
Also in 1988, we spent about 1,400 hours on a glorified name and address system
to keep track of the various classes of business partners that we have. That
system was a disaster (also, thankfully, before my tenure began). In 1992, we
spent well over 1,600 hours rewriting a "new and improved" version. In 1995, a
very good C++ programer began rewriting that system again. About two months
into the project, we started looking at VSE. After about another month, we
abandoned our attempt to use C++, and focused on Smalltalk. We did not persue
that particular system at that time, as we were new to Smalltalk and didn't
want to risk another failed attempt at that system. We selected two other
projects, and both were solid success stories. In 1998, the name and address
system was rewritten in Smalltalk in less than 400 hours. In 1988, we had 8
developers hacking away. In 2000, we have 2, yet we actually produce more new
software each year than we did in 1998 - 1992.
When VSE died (or was killed), things looked a little grim. I based my
recommendation to switch to VAST on the experience I had with the language.
Since I had the numbers, and since I had been able to complete two systems in
VSE, I could show the difference. We bet our Y2K remediation effort on VAST,
and we won!
I'd also like to point out the following:
Most popular car in 1999 - Ford Taurus
http://www.auto.com/reviews/topreviews.htm
Highest Grossing Movie in 1999 - Star Wars The Phantom Menace
http://movieweb.com/movie/alltime.html
Top retail software title 1994- 1998 - Windows 95 upgrade
http://www.infoplease.com/ipa/A0777825.html
The top rated show, (actually 7 of the top 10 shows:1/10/00 - 1/16/00)
MILLIONAIRE (Who wants to be....)
http://www.ultimatetv.com/news/nielsen/networks/000110network.html
Popular? Yes, very much so. Best? Well, you be the judge As for me, I'll
continue to explore and learn JAVA. I still have a C++ compiler. But as long
as I can, I'll develop in Smalltalk.
Dan Antion
American Nuclear Insurers
Taylor Corey wrote:
Frank Sergeant
> Er...Not yet! But if a mere M.A. *in* philosophy will do, I'm at your
> service :)
Thank you. The advice was very welcome.
> > I had a nice talk on the phone with
> > Cincom about two weeks ago and had the different options explained
> > to me.
>
> Hey! Explain 'em to us! :)
I spoke with Richard Woyce in Atlanta at (770) 441-0550 x213. He
was very patient and spoke with me at length. I had probably hinted,
in previous postings, that I had become rather annoyed, not just that
I might not like the new price of VW, but that I couldn't even find
out what that price was! I think Cincom should say plainly what the
price is, and post that information prominently on its web site.
Apparently Cincom disagrees with me on this point. Someone suggested
it was Cincom's policy not to post its prices publicly. This forces
a prospect to contact a salesman (or to delay doing so or to look
into alternatives). In spite of my disagreeing with Cincom's apparent
policy, if they do not wish to have the prices published then I'm
reluctant (there's that word again) to publish them. I'll be happy to
discuss any of this in private email, however. I invite each of you who
might be interested to give Richard (or other Cincom contacts) a call.
> > ... I have a reluctance to agree to paying a royalty.
> As do many. But if you're accounting can handle it without much
> difficulty I wouldn't think this should be a dealkiller.
Good point. I'd have to talk further with Cincom to find out how
they would want to handle the accounting, just what the royalty
would be based upon, etc. We offer our customers an annual support
service which includes program upgrades plus telephone (and fax and
email) support. Should the royalty be on our total income or just
on the percentage "allocated" to product upgrades? When we sell
a program, do we pay a royalty on the entire amount (we often spend
considerable time installing and training, if a salesman is involved,
we pay a substantial commission) or on the amount after subtracting
amounts "allocated" for installation, training, sales commission.
Before reaching this point in my thinking, I had turned to seriously
considering alternatives to VW.
> (But *note* Cincom folks! It *is* a barrier. Even people *willing* to go
> along with it feel *reluctance*!
It _is_ a barrier, to me at least, whether I am being rational or not!
A question is whether I am sufficiently typical for Cincom to give
any weight to my thoughts on the subject.
I had grown fond of VW. I was prepared to pay the $500 price and
consider later paying the $3000 price (when those were thought to
be the prices of personal and enterprise versions, prior to Cincom's
price policy changes). It was announced that the price policy was
changing and that it was likely to be royalty based. I didn't like
that, failed to get answers about the price by email and failed to
get a call back by telephone. Meanwhile, I put my VW work on hold
and began working with some alternatives. Finally, out of curiosity
before abandoning VW altogether, I phoned again, and had the pleasant
conversation with Richard that I mentioned. Well, now I'm developing
inertia in another direction although I see pros and cons to VW and
am still considering it.
> You don't want that! Make it easy. If
> you're going to have royalties, make 'em clear as day! Look at the
> licencing of Macintosh Common Lisp (MCL) http://www.digitool.com. They
> seem quite reasonable, and unconfusing. Don't call them royalties, call
> them "distribution" rights and limit them to a dollar amount, number,
> geographic region, organization, or the like.)
Thanks. I am tempted to switch from Windows/Linux on Intel machines
and use a Mac so I can use MCL (no, not really), but I did like the
site. At least Digitool spoke right up and announced their prices
and I could use their Lisp without paying royalties.
> Very very much depends on what you're doing and why, but my personal
> basic thought is: If your task just *screams* for VisualWorks, and the
> pricing structure doesn't kill the deal, then go for it. Dithering is
> likely to be as wasteful as plunging, and it's a lot more painful.
Thanks. I'm still dithering and it is painful. I don't see that the
project *screams* for VW.
> (Personally, I think Cincom should divorce licence fees and tech
> support, and be liberal with the licences. You needn't GPL something to
> make your money of tech support.
I'd like to see that as well.
> But, in general, if they are going to be your partner, hold them to it.
> Think about what you'd expect and want from a partner, and see if
> they'll do that.
Good point. There more I think about it, the more I think I don't
_want_ any more partners.
Thanks again, very much, for the advice.
-- Frank
fr...@canyon-medical.com
Taylor Corey
to use Gemstone or a traditional database for their ST front-end. Several
of us recommended Gemstone because the development time would be much
shorter and the final product more robust. A VP, who attended the meeting,
said he can not win that argument in his company. He said the way American
business works is, 'How much money did I save you today, not 6 months from
now.'.
If a heating/air conditioning guy came to your house and said, 'You have two
choices for a heater. Option 1 costs less but is less efficient. Option 2
costs more but you will earn back the extra money paid over the course of
two years because the system is much more efficient and your heating bills
will be much lower.'.
If your were a business, then most likely, you would choose option option 1.
If you were a private individual then you might choose option 2. That's not
a hard and fast rule but in most cases, business look to save money today.
--tc
Dan Antion <ant...@attglobal.net> wrote in message
news:388E3DF3...@attglobal.net...
Taylor Corey
31K. I think it would be fair to say that the 'ST Report' did not even come
close to that. The 'ST Report' would talk about how their numbers were
growing but never gave out actual figures. That is one of the problems with
the ST industry. People like to talk a good game but play it very close to
the vest when it comes time to back their claims up with numbers.
Java and other languages have several magazines devoted exclusively to them,
ST does not even have one.
--tc
Eric Rizzo <eri...@tampabay.rr.com> wrote in message
news:388E53AE...@tampabay.rr.com...
> Taylor Corey wrote:
> >
> <snip>
> > Here are some numbers:
> >
> <snip>
> > Magazines
> > ========
> > 1. Smalltalk,
> > - An article in JOOP.
> > - St Report went out of circulation. In its best year, I don't think it
> > came close to 70K subscribers but someone correct me if that is wrong.
> > 2. Java,
> > - 'Java Developer's Journal' has a subscriber base of 70K,
> > http://www2.sys-con.com/recordcirculation.cfm
>
> Sure, if I send out 70K copies of my "Manure-based Electricity
> Solutions" newsletter to people who pay nothing for it and are as likely
> use it to line their bird cages as they are to actually read it, then I
> can claim that Manure-based electricity solutions are more popular than
> traditional ones.
> The deal with JDJ, to which I "subscribe," is that they give it away to
> anyone whose address they can get their hands on, and include far more
> add pages than actual content to subsidize this practice. Claiming more
> than twice the "circulation" than Java Report, which I estimate most
> Java developers would rate a publication of higher overall quality?
> Come on, get real.
>
> Point is, JDJ is not an accurate measure of Java popularity, momentum,
> quality, etc. Note I am not arguing those issues, just the use of JDJ
> as an example.
>
> Eric
>
> --
> Eric Nicholas Rizzo
> eri...@tampabay.rr.com
> ---------------------------------
> "Don't make me come down there!"
> -God
Taylor Corey
might buy a less popular, newer model car which may have advantages but also
might cost more, and be harder and more expensive to service. On the other
hand, you could buy Taurus' which may have disadvantages but if something
goes wrong, your butt won't be as exposed to upper management, IMHO.
--tc
> > Magazines
> > ========
> > 1. Smalltalk,
> > - An article in JOOP.
> > - St Report went out of circulation. In its best year, I don't think it
> > came close to 70K subscribers but someone correct me if that is wrong.
> > 2. Java,
> > - 'Java Developer's Journal' has a subscriber base of 70K,
> > http://www2.sys-con.com/recordcirculation.cfm
> >
Eric Rizzo
>
<snip>
> Here are some numbers:
>
<snip>
> ========
> 1. Smalltalk,
> - An article in JOOP.
> - St Report went out of circulation. In its best year, I don't think it
> came close to 70K subscribers but someone correct me if that is wrong.
> 2. Java,
> - 'Java Developer's Journal' has a subscriber base of 70K,
> http://www2.sys-con.com/recordcirculation.cfm
Sure, if I send out 70K copies of my "Manure-based Electricity
Dan Antion
again, I didn't exactly make it, I only implied it. The part of my message I'm
referencing is "in 1988 there were 8 developers..... today there are two". It's
not fun, it's not something anybody wants to say, particularly most IT managers,
but you can work in Smalltalk with fewer developers than other languages,
period. We didn't go from 8 to 2 when we selected Smalltalk, but we did go from
6 to 3 and then when one other one left, there we made no argument to replace
him.
Dan
James A. Robertson
>
> I have the same question. How does one become an effective advocate for ST?
> Smalltalk is a niche market and some people/companies do not want to play in
> that arena. I said before the number of VAST licenses totals 80K, I'll
> retract that because I think the real number is more like 8K.
>
All I can tell you is that activity in this group is much higher, and
includes a lot of new people since 1995. Most likely Eric Clayberg has
a better feel for VAST licenses than I do.
--
James A. Robertson
Senior Sales Engineer
Cincom Systems, Inc.
jar...@mail.com
Talk Small and Carry a Big Class Library
James A. Robertson
>
> I'm speaking commercially. Of course, the big ST powerhouse used to be HBO
> Medical. They have since abandoned all st development.
Not really true. They made a decision like that two years ago. In
light of that, many of their ST developers bailed. They <then>
discovered how much of their infrastructure depended on Smalltalk.
As of last year, they were still paying for support for VW.
James A. Robertson
>
> What you say may be true. I was trying to be generous in my use of numbers
> but if software licenses are a smaller part of the IT budget as you say,
> that VAST licenses are below OBJS' peak of 6K - 8K but they wouldn't be much
> more either.
>
Why not ? IBM had a <much> larger sales force pushing product than OBJS
ever had. They made a decent business merely by introducing VAST as a
way of going from Cobol --> OO.
> --tc
>
> "DOUGLAS A SWARTZ" <DASW...@prodigy.net> wrote in message
> news:86hkbi$sk0$1...@news.software.ibm.com...
> > Software licenses are a tiny portion of the overall IT budget. The IT
> budget
> > includes hardware, all software developer and consulting salaries, system
> > administration and other costs. I think you're trying to compare apples
> and
> > oranges.
> >
> > Doug Swartz
> > dasw...@prodigy.net
> >
> >
--
Taylor Corey
up to exit 11 on GA 400 (Windward Parkway) and there is no ST going on in
their new building.
--tc
James A. Robertson <jar...@home.com> wrote in message
news:388F95A6...@home.com...
> Taylor Corey wrote:
> >
> > I'm speaking commercially. Of course, the big ST powerhouse used to be
HBO
> > Medical. They have since abandoned all st development.
>
> Not really true. They made a decision like that two years ago. In
> light of that, many of their ST developers bailed. They <then>
> discovered how much of their infrastructure depended on Smalltalk.
>
> As of last year, they were still paying for support for VW.
>
>
Taylor Corey
software license revenues. If ST were more popular than one would think
people would be interested in having at least one ST magazine.
--tc
James A. Robertson <jar...@home.com> wrote in message
news:388F96B3...@home.com...
Taylor Corey
going but no one seems to know where it is right now. The people who do
know won't talk about it. However, if ST were as popular as some claim then
one would think other evidence of its popularity, like an ST magazine, would
exist.
The Gartner Group said ST will represent less than 5% of all OO activity
thru 2004. Another indicator, IBM has appointed a VP of Java. If ST is
popular than where is the VP of Smalltalk?
Here is a snippet from PC Weak on OO at
http://www.zdnet.com/pcweek/stories/news/0,4153,2381666,00.html :
There is a difference between making the right decision now and
second-guessing decisions that were made years ago. For example, Smalltalk
is an absolutely wonderful programming language--it is the first
object-oriented language to gain popularity in the business programming
community, and there are many applications using it today. If we committed
to a project using Smalltalk in 1992, we would have been taking less risk
than using other object-oriented languages available at the time.
But few people in 1999 would commit a major new development project to that
language. For any number of reasons, Java presents less risk to the
organization. The primary risk we were taking in 1992 was that object
orientation would catch on--it has--and that Smalltalk would be its major
implementation environment--it isn't. Win one, lose one.
Bijan Parsia
> This gets back to my previous post. People love to tell others where ST is
> going but no one seems to know where it is right now. The people who do
> know won't talk about it. However, if ST were as popular as some claim then
> one would think other evidence of its popularity, like an ST magazine, would
> exist.
Well, do goodstart.com and whysmalltalk.com, and the object people
online magazine count?
I think it's really really hard to guage the *lack* of "popularity" of a
langauge by the bookstore shelf/magazine rack space measure. (Actually,
positive popularity is tricky too.) After all, these are *secondary* (or
even tertiary) markets with their own dynamics. Plus, they tend to lag
behind "market" trends---to wit, there was a bit of a collapse of
Macintosh mags just as Apple was turning around.
Now, granted, lack of mags or books (on shelves) tends to have a bit of
negative *causal* effect, but that, too, isn't determinative.
Plus, the typical *useful* question is, "Is Smalltalk 'popular' enough
to be a good choice for me for task foo?" And frankly, there's something
a bit weird about that question. (Of course, there's the worry about
venders and finding people, etc., but I haven't noticed that that's a
killer objection for Smalltalk. The major commerical variants seem
relatively stable (*VW* seems likely to persist, regardless of vendor).
There are Free/Open Source versions. There's an ANSI standard. There are
more independant (at least, "from scratch") active implementations, I
think, than ever before.
The "standard" answers to the "can I find programmers" question is,
"Yes", and "It's easier to train people than you might think." Obviously
these answers won't work for everybody.
> The Gartner Group said ST will represent less than 5% of all OO activity
> thru 2004.
BWAHAHAHAHA! 1) What do they say it represents now? 2) "OO activity"?
What's that? New development? Programmers working or hired? LOC in play?
Number of programs sold? 3) How much is "OO activity" going to expand
"thru 2004"? 4) Is *all* OO activity the right measure? What about
various niches? 5) Etc.
(These are just the questions *favoring* Smalltalk. What if "OO
activity" (gotta love that) shrinks? Etc.)
And this is *totally* putting aside questions of methodology. I'm
*extremely* skeptical about claims about "activity" of a computer tech
"thru 2004".
And, frankly, the Gartner Group is not my personal idea of a reliable
source. Alas, they *are* taken seriously, so they may have an effect.
But that's immoral, IMHO :)
> Another indicator, IBM has appointed a VP of Java. If ST is
> popular than where is the VP of Smalltalk?
"More popular". You mean "more popular than before"? Or "more popular
than Java"? I hope no one claims the latter.
Maybe they don't need or want a VP of ST. But that could be for any
number of reasons, yes? Politics, size of department, problems in a
department, etc. etc.
> Here is a snippet from PC Weak on OO at
> http://www.zdnet.com/pcweek/stories/news/0,4153,2381666,00.html :
>
> There is a difference between making the right decision now and
> second-guessing decisions that were made years ago. For example, Smalltalk
> is an absolutely wonderful programming language--it is the first
> object-oriented language to gain popularity in the business programming
> community, and there are many applications using it today. If we committed
> to a project using Smalltalk in 1992, we would have been taking less risk
> than using other object-oriented languages available at the time.
This may sound good, but it's empty. How on Earth is "risk" being
calculated. Indeed, what is it risk *of*. Of ending up using a less
popular programming langauge? Gee, why should one care? (Obviously, that
Smalltalk would disappear, or have no viable venders, or have no Open
Source varients all turned out ok.)
> But few people in 1999 would commit a major new development project to that
> language. For any number of reasons, Java presents less risk to the
> organization.
Really? Than Smalltalk? For what? How's this measured?
It also totally ignores that committing to some particular Java vender
or tech might be *very* "risky".
> The primary risk we were taking in 1992 was that object
> orientation would catch on--it has--and that Smalltalk would be its major
> implementation environment--it isn't. Win one, lose one.
I don't understand why a *user* of Smalltalk faces either of these
risks. If using Smalltalk at the time was effective in producing profit,
then who cares if *either* object orientation caught on *or* Smalltalk
was it's major "implementation environment" (I don't even *know* what
that's supposed to mean. Sounds good, less filling.)
If you are a Smalltalk *vendor* with certain *goals* those goals may
have been or be at risk due to Java. Being the major vm bundled with web
browsers in 2000, that's a goal which is "at high risk" (to put it
mildly) of failure (for a non-java compatible vm), especially if one
isn't Microsoft. For Microsoft, of course, it's an (almost) no-risk of
failure goal (though it might cause *other* problems).
I don't know why people worry about "risk" and try to make "less risky"
decisions if they aren't going to take the care to get reasonable
understandings and evalutions of the risk involved. Flip a coin!
*That's* less risky, less time and angst wasting, and less silly.
(Ach! Lets not even get into perceived risk and *it's* secondary effects
:))
Just so this doesn't...er...degenerate (further? :)), we could try to
figure out a reasonable decision tree, or do a case study, or try to
dicuss what would count as good evidence, etc. Let's *not*, until those
discussions happen, debate "where smalltalk is or where it's going".
Cheers,
Bijan Parsia.
James A. Robertson
> Taylor Corey <tco...@mindspring.com> wrote:
>
> > This gets back to my previous post. People love to tell others where ST is
> > going but no one seems to know where it is right now. The people who do
> > know won't talk about it. However, if ST were as popular as some claim then
> > one would think other evidence of its popularity, like an ST magazine, would
> > exist.
>
I also think it's a bit dangerous to draw a complete conclusion based on
one data point. Your example (HBO&C) had been after ParcPlace since at
least 1994 for native widgets and better COM support. Had ParcPlace not
killed the Van Gogh project, they most likely would have stayed happy.
Taylor Corey
> I also think it's a bit dangerous to draw a complete conclusion based on
> one data point.
I agree and the question still remains, 'Where is smalltalk now?'. When I
say where is smalltalk, that means how many dev licences are out there
compared to other languages like C, C++, Java, etc. It is important to keep
things in context. The people who know won't talk. The Gartner Report on
license revenue is several years old and costs like $5K to purchase. The
only thing left is to resort to public sources.
We basically get one of the following posts on the forum:
1. Big things are happening with ST.
2. [some name] doesn't know what they are talking about.
3. Who cares.
4. It's all FUD anyway.
It is a little hard to know where you are going if you don't know where you
are now. Using some of the web pages as sources then one could say there
are 45(?) companies with active projects. There is no ammo at all to use
for convincing people within a company to use ST. The ST web pages do not
have context info which is important for managers who need to consider
alternatives. The only thing I've seen that puts ST into context is the
Gartner report that says ST development will represent less than 5% of all
OO activities thru 2004.
Maybe STIC should have a link that says, 'Where is smalltalk now', and have
a market share report from the Meta Group, Gartner Group, etc.
--tc
Bijan Parsia
[snip]
> I agree and the question still remains, 'Where is smalltalk now?'. When I
> say where is smalltalk, that means how many dev licences are out there
> compared to other languages like C, C++, Java, etc.
But isn't this totally unmeasureable, since there really isn't anyway to
track (espeically) C and C++? (I mean, every Linux comes with GCC.)
They're ubiquitous and often necessary for non development work
(compiling distributed code, for example).
Similar difficulties for, say, Squeak. And free Dolphin, though, there,
we can perhaps track downloads. That would be *some* data.
But, hey, let's put aside methodological issues. What is the best *spin*
to put out for Smalltalk? ;)
> It is important to keep
> things in context. The people who know won't talk. The Gartner Report on
> license revenue is several years old and costs like $5K to purchase. The
> only thing left is to resort to public sources.
Er...Ok.
(I'm trying to get myself into the mind of someone in a company trying
to sell ST to a company. <shrug/> That's not my situation, so I'll have
a little trouble :))
[snip]
> It is a little hard to know where you are going if you don't know where you
> are now. Using some of the web pages as sources then one could say there
> are 45(?) companies with active projects. There is no ammo at all to use
> for convincing people within a company to use ST.
I guess not...if the way one argues is by "raw" numbers. But if one of
those companies is doing something similar to what you need to do, isn't
*that* relevant? Espeically if the company has a good rep.
Erlang seems to do well, since Ericssan can point to their own use of
it. Smalltalk has something a bit odder working "against" it.
> The ST web pages do not
> have context info which is important for managers who need to consider
> alternatives.
Context being what I mentioned above?
> The only thing I've seen that puts ST into context is the
> Gartner report that says ST development will represent less than 5% of all
> OO activities thru 2004.
Oh, not. But ST isn't going to win the numbers game. Unless we point to
Redhat's distributing VWNC for Linux :)
> Maybe STIC should have a link that says, 'Where is smalltalk now', and have
> a market share report from the Meta Group, Gartner Group, etc.
I guess. Or more case studies, whatever.
I don't know if anyone can win over people who assess risk based
entirely on popularity. But it's not at all clear that that's needed for
(various sorts of) success as well.
Cheers,
Bijan Parsia.
Richard MacDonald
news:1e522xl.1yqfmhocz94ccN%bpa...@email.unc.edu...
> >
> > There is a difference between making the right decision now and
> > second-guessing decisions that were made years ago. For example,
Smalltalk
> > is an absolutely wonderful programming language--it is the first
> > object-oriented language to gain popularity in the business programming
> > community, and there are many applications using it today. If we
committed
> > to a project using Smalltalk in 1992, we would have been taking less
risk
> > than using other object-oriented languages available at the time.
>
> This may sound good, but it's empty. How on Earth is "risk" being
> calculated. Indeed, what is it risk *of*. Of ending up using a less
> popular programming langauge? Gee, why should one care? (Obviously, that
> Smalltalk would disappear, or have no viable venders, or have no Open
> Source varients all turned out ok.)
Risk factors: Picking the wrong Smalltalk platform, i.e., VSE.
Inability to find good programmers. Need to double their salary to
keep them away from Sprint. (And yes, you can put the last two together
and conclude that I meant "inability to find good programmers willing to
settle for the wages I can get passed through my business political
minefield." In other words, risk that the dysfunctional portion of my
company wins over the common-sense portion.)
Empty my ass. The risk formula may be difficult to calculate, but I'm all
too
familiar with its effects, thank you very much.
> > But few people in 1999 would commit a major new development project to
that
> > language. For any number of reasons, Java presents less risk to the
> > organization.
>
> Really? Than Smalltalk? For what? How's this measured?
And from the other side of my mouth, my company *is* committing development
funds to Smalltalk. We're still only committing to my hours and not
purchasing
additional tools, but with a little proven success, that will come. (And if
the
oil companies start spending again. Am I the only person in this newsgroup
who cheers when the price of oil rises :-?)
> It also totally ignores that committing to some particular Java vender
> or tech might be *very* "risky".
IMHO, *more* risky than the Smalltalk risk.
> I don't know why people worry about "risk" and try to make "less risky"
> decisions if they aren't going to take the care to get reasonable
> understandings and evalutions of the risk involved. Flip a coin!
> *That's* less risky, less time and angst wasting, and less silly.
I understand your point. But I would go further and say its *all* risk.
Look into the crystal ball. Any decision made today will be wrong in
3 months. You can develop any risk formula you want and it'll be bogus.
You *are* flipping a coin.
Taylor Corey
> [. . . snip . . .]
> But isn't this totally unmeasureable, since there really isn't anyway to
> track (espeically) C and C++? (I mean, every Linux comes with GCC.)
> They're ubiquitous and often necessary for non development work
> (compiling distributed code, for example).
Typically, what a Gartner Group type company would do is take a statistical
sample of companies and survey them asking the number of projects, what
languages are used, etc.
> [. . . snip . . .]
> I don't know if anyone can win over people who assess risk based
> entirely on popularity. But it's not at all clear that that's needed for
> (various sorts of) success as well.
I don't necessarily agree with how managers evaluate whether to use
something or not. One thing I fairly confident of is most managers are not
going to stick their butt out needlessly. They have to be given something
whether it is popularity, your competitor is doing it, etc. No one thing
alone would win the battle but it would help if one had several major arrows
in their quiver.
I can't hardly use the argument of, ' trust me, st is better' to a manager.
Especially if IBM is pushing Java.
This puts us back at square on, we don't have have a context for st and
hardly no ammo at all for talking to upper management about it.
--tc
Thierry Thelliez
> This puts us back at square on, we don't have have a context for st and
> hardly no ammo at all for talking to upper management about it.
>
> --tc
Yep....
Althought I did score 1 point this week when I gave the story of soliloquy
startup to my
management (thanks to whoever forwarded that!).
Do we have any numbers about the current ST growth ? Few years ago, ParcPlace
announced
a 250% growth. What is the situation today ? How IBM, Cincom, GemStone are
doing in the ST market ?
GemStone/J seems to be growing quite fast compared to GemStone/S, but there has
been 0
marketing for GemStone/S.
Thierry
Thierry Thelliez
But what about conferences ? I just got the agenda of Software Development
Conference
www.sdexpo.com and there is not a single track about ST. There will be C++, Java,
Python, XP, COM/Corba, XML, App servers, Beans, JDBC, OR mapping,...
This drives me really mad.... Where is ST advertised ? Where is ST
publicly presented ? How can we wake up ST marketing people ?
Thierry
John Clonts
>
> > ..
> > This puts us back at square on, we don't have have a context for st and
> > hardly no ammo at all for talking to upper management about it.
> >
> > --tc
>
> Yep....
>
> Althought I did score 1 point this week when I gave the story of soliloquy
> startup to my
> management (thanks to whoever forwarded that!).
Huh? Did I miss something? Please elaborate :)
Cheers,
John
Bijan Parsia
> Bijan Parsia <bpa...@email.unc.edu> wrote in message
> news:1e522xl.1yqfmhocz94ccN%bpa...@email.unc.edu...
> > >
> > > There is a difference between making the right decision now and
> > > second-guessing decisions that were made years ago. For example,
> Smalltalk
> > > is an absolutely wonderful programming language--it is the first
> > > object-oriented language to gain popularity in the business programming
> > > community, and there are many applications using it today. If we
> committed
> > > to a project using Smalltalk in 1992, we would have been taking less
> risk
> > > than using other object-oriented languages available at the time.
> >
> > This may sound good, but it's empty. How on Earth is "risk" being
> > calculated. Indeed, what is it risk *of*. Of ending up using a less
> > popular programming langauge? Gee, why should one care? (Obviously, that
> > Smalltalk would disappear, or have no viable venders, or have no Open
> > Source varients all turned out ok.)
>
> Risk factors: Picking the wrong Smalltalk platform, i.e., VSE.
But not a risk of Smalltalk (in general) over anything else. Well, maybe
slightly over Java *if* you think Java is more vender portable.
> Inability to find good programmers. Need to double their salary to
> keep them away from Sprint.
Absolutely. Of course, the difficulty of finding *good* programmers is
ubiquitous. And good & cheap is even harder.
> (And yes, you can put the last two together
> and conclude that I meant "inability to find good programmers willing to
> settle for the wages I can get passed through my business political
> minefield." In other words, risk that the dysfunctional portion of my
> company wins over the common-sense portion.)
Er.. these are fine. These are real. But I was really critiquing the
particular article where "risk" seemed to mean "risk of not programming
in the most popular langauge". When I could nail it down at all.
> Empty my ass. The risk formula may be difficult to calculate, but I'm all
> too
> familiar with its effects, thank you very much.
Er...I hope you had a comma after the 'empty' up there :)
I think there are real risk formulas, but they tend to be specific (and
even then, they are difficult to calculate). And you have to get the
"outcome" right (i.e., is "not programming in the most popular langauge"
in and of itself a *bad* outcome? in what way (granting possible
secondary effects like vender stability and programmer availability, but
that way complicates the formula since you have to calculate the
*dependant* risk).
Again, I wasn't arguing against figuring risk, or even, for many areas,
going with Smalltalk might be much riskier.
> > > But few people in 1999 would commit a major new development project to
> that
> > > language. For any number of reasons, Java presents less risk to the
> > > organization.
> >
> > Really? Than Smalltalk? For what? How's this measured?
(See see. It's specific to this article! For being the dominant
in-browser applet delivery mechanism, Smalltalk is *way* risker than
Java :) Of course, *as* a in-browser applet delivery mechanism *for
various audiences, it might have a rather low risk of failure.)
(I objected and object to talking about risk as if it were completely
divorced from situation and goals and had a direct link to popularity.)
> And from the other side of my mouth, my company *is* committing development
> funds to Smalltalk. We're still only committing to my hours and not
> purchasing
> additional tools,
Hey, that's something!
> but with a little proven success, that will come.
Cool.
> (And if
> the
> oil companies start spending again. Am I the only person in this newsgroup
> who cheers when the price of oil rises :-?)
No! I think there should be a huge tax on oil and gas! (Not that this
would help the oil companies :))
> > It also totally ignores that committing to some particular Java vender
> > or tech might be *very* "risky".
>
> IMHO, *more* risky than the Smalltalk risk.
Well, even this is difficult. It depends on what you do. If you go in
heavily for training and very vender specific extentions and facilities,
then you end up very tied to the success of the vender (unless, of
course, they open source, or liberal licence, or ...), but if you go
"pure Java", your product might be too slow, or underpowered, or...
And even if your vender goes under and that's a problem, there's the
question of how easy it is to recover, etc. etc.
I think people go to business school to learn how to figure this out.
Too bad they don't write magazine articles ;)
> > I don't know why people worry about "risk" and try to make "less risky"
> > decisions if they aren't going to take the care to get reasonable
> > understandings and evalutions of the risk involved. Flip a coin!
> > *That's* less risky, less time and angst wasting, and less silly.
>
> I understand your point. But I would go further and say its *all* risk.
> Look into the crystal ball. Any decision made today will be wrong in
> 3 months. You can develop any risk formula you want and it'll be bogus.
> You *are* flipping a coin.
Embrace change :)
Ah, I should have gone further too :) I didn't want to scare folks off
;) I do think you can get some useful data about some things. But I'm
not sure that typically the effort to get good data is worth the cost.
(Another neglected point: Can you make a good risk assessment for less
that it would cost you if you choose wrong? Espeically if you invest
that effort & resources into prepping for the various cases. eXtreme
Management :))
Cheers,
Bijan Parsia.
Richard MacDonald
>
> > > Empty my ass. [snip]
> >
> > Er...I hope you had a comma after the 'empty' up there :)
>
>
Harrumph. I was hoping no one else would notice :-)
Doug Way
I agree with your objection to this usage of "risk".
Is language popularity really the *only* consideration when it comes to
the risk involved with choosing a language? It seems that the Gartner
group would have us think so, as this is what their studies appear to be
based on.
I would think the biggest risk involved would be project failure. Or at
least the risk of having a project take too long/cost too much, as well
as the cost of maintenance over its lifecycle. *This* is the real
"risk" value. The smaller (but still significant) risk of using an
unpopular language would factor into this value, by virtue of the
difficulty involved with finding good programmers to complete the
project, etc.
Does the Gartner group have any studies which look at success rates for
projects using different languages? Or do they study the ratio of how
much projects cost, given the complexity of the application (in function
points)? (Maybe they do, I'm not sure... if not, hopefully someone else
does.) These types of studies might actually be worth something, as
opposed to language popularity studies. (Although I admit these types
of studies involve a lot more effort.)
I tend to think that some managers use language popularity as the main
measure of risk, simply because it is the easiest factor to measure.
If some good studies indicated that Java had a significantly higher or
even similar project success rate compared with Smalltalk (which I
doubt), I would then have no problem going with Java.
Also, I heard a figure of Smalltalk having a 5-10% market share in OO
development. Is that really so low? I certainly wouldn't be concerned
with buying, say, a Chrysler car, even though their market share is
similar. Apple seems to be doing fine with this market share. You
could even argue that such a market share is self-sustaining (as opposed
to dwindling), given that some people will always want some sort of
alternative to the mainstream, if they perceive it as superior. When
you get under 1%, I would start to become concerned.
> > Empty my ass. The risk formula may be difficult to calculate, but
I'm all
> > too
> > familiar with its effects, thank you very much.
>
> Er...I hope you had a comma after the 'empty' up there :)
(Took me a minute to get this one. Hee!)
> Ah, I should have gone further too :) I didn't want to scare folks off
> ;) I do think you can get some useful data about some things. But I'm
> not sure that typically the effort to get good data is worth the cost.
Good point. It is probably too expensive for a company doing their own
research. But I would think a consulting group (e.g. Gartner)
specializing in this would be able to afford some good studies...
- Doug Way
EAI/Transom Technologies, Ann Arbor, MI
http://www.transom.com
dw...@mat.net, @eai.com
Sent via Deja.com http://www.deja.com/
Before you buy.
Dan Antion
needs, and realized I could meet those needs with a smaller fleet of slightly
more expensive mini-vans? As I said in my other reply, the point that may be
missed in my original reply is that we are developing more applications today
with fewer programmers.
Management wants a reason to select a language. Many people in this forum agree
that developing in Smalltalk is more productive. That translates (or can
translate) into fewer bodies, lower overall cost. that worked for me. We did
select Smalltalk, we do have fewer programmers, and we are getting the job
done. If you have 20 programmers and you want to keep 20 programmers, and your
development need and time to market is fixed, then it's hard to argue for any
productivity improvement. If you want to get the work completed faster, or
completed at an overall lower cost, then you can make a very good argument for
Smalltalk.
Dan Antion
Taylor Corey wrote:
> If you were charged with buying a fleet of cars for your company then you
> might buy a less popular, newer model car which may have advantages but also
> might cost more, and be harder and more expensive to service. On the other
> hand, you could buy Taurus' which may have disadvantages but if something
> goes wrong, your butt won't be as exposed to upper management, IMHO.
>
Bijan Parsia
> In article <1e532z7.1lcg7gpnq0nx6N%bpa...@email.unc.edu>,
> bpa...@email.unc.edu (Bijan Parsia) wrote:
[snip]
> > Er.. these are fine. These are real. But I was really critiquing the
> > particular article where "risk" seemed to mean "risk of not
> programming
> > in the most popular langauge". When I could nail it down at all.
>
> I agree with your objection to this usage of "risk".
[snip]
> I would think the biggest risk involved would be project failure.
Perhaps "most significant" risk. If your project fails, it's no comfort
that *at least* it was written in the most popular language. At least,
most of the time. It may be the case that you can recover something by
spinning off parts of the project (libraries, etc.).
> Or at
> least the risk of having a project take too long/cost too much, as well
> as the cost of maintenance over its lifecycle.
Right on. Popularity of the langauge *can* be an indicator, via its
being a (very rough) indicator of other things which *can* affect the
direct success of a project. But, whoa, the connection is really quite
indirect.
Unless, of course, your knowledge is particularly sparse and you want to
rule out altogether certain risks so you can focus on others. (So, if
you go with Java, you can totally neglect the fear that there might not
be a Java book on your local bookstore's shelves. Whew! :)) But then you
(can) end up in various prisoner dilemma situations.
So, for example, abstractly we can be reasonably sure that there are
more Java programmers than Smalltalk programmers. But if I thereby
decide to rule out Smalltalk, I might well miss that there is a *slew*
of programmers *expert* in my problem domain who vastly prefer working
in Smalltalk, and who would, perhaps, take a lower salary, or be worth
more, etc.
(Obviously, the reverse can be true. I'm just trying to illustrate that
these decisions *oughten* to be made on just wifty analysis, as the
article suggested.)
> *This* is the real
> "risk" value. The smaller (but still significant) risk of using an
> unpopular language would factor into this value, by virtue of the
> difficulty involved with finding good programmers to complete the
> project, etc.
Right, taking into account the *specifics* of each situation.
> Does the Gartner group have any studies which look at success rates for
> projects using different languages? Or do they study the ratio of how
> much projects cost, given the complexity of the application (in function
> points)? (Maybe they do, I'm not sure... if not, hopefully someone else
> does.) These types of studies might actually be worth something, as
> opposed to language popularity studies. (Although I admit these types
> of studies involve a lot more effort.)
Dunno. I will say I'm not generally impressed with what I've seen of the
GartnerGroup. In my circles, they tend to be snarled about ;)
> I tend to think that some managers use language popularity as the main
> measure of risk, simply because it is the easiest factor to measure.
Oh sure. And sometimes, that's the right thing to do. But I don't think
that we should encourage this as a general phenomenon.
> If some good studies indicated that Java had a significantly higher or
> even similar project success rate compared with Smalltalk (which I
> doubt), I would then have no problem going with Java.
Before going for it, make sure the comparisons make sense for your
situation. There are a *lot* of factors that can affect project success,
and I suspect that they dominate language choice in most cases.
> Also, I heard a figure of Smalltalk having a 5-10% market share in OO
> development. Is that really so low? I certainly wouldn't be concerned
> with buying, say, a Chrysler car, even though their market share is
> similar. Apple seems to be doing fine with this market share. You
> could even argue that such a market share is self-sustaining (as opposed
> to dwindling), given that some people will always want some sort of
> alternative to the mainstream, if they perceive it as superior. When
> you get under 1%, I would start to become concerned.
Er...why? Market share alone doesn't determine market *size*. Would you
rather have 1% of a 10inch pie, or 25% of a 2inch pie :)
Like popularity arguments, market share (which is intimately connection
with popularity, of course) arguments rely mostly on secondary effects,
fears of monopoly positions, etc.
Which are real effects, of course, but the "oh you should program for
Windows since you'll have a bigger market" tended to ignore the fact
that for many products *you* would have a "bigger" market in the Mac
world.
> > > Empty my ass. The risk formula may be difficult to calculate, but
> I'm all
> > > too
> > > familiar with its effects, thank you very much.
> >
> > Er...I hope you had a comma after the 'empty' up there :)
>
> (Took me a minute to get this one. Hee!)
It took me a minute to read it the *right* way, I'm afraid. I was rather
taken aback :) Shows you where *my* mind is.
> > Ah, I should have gone further too :) I didn't want to scare folks off
> > ;) I do think you can get some useful data about some things. But I'm
> > not sure that typically the effort to get good data is worth the cost.
>
> Good point. It is probably too expensive for a company doing their own
> research. But I would think a consulting group (e.g. Gartner)
> specializing in this would be able to afford some good studies...
*Can* afford, probably. *Make stupendous profit off of*, less likely :)
I do wish there were funds to do neutral studies of Smalltalk. Not
necessarily comparative, though those are interesting too. Oh well. I
program for fun and personal reasons anyway :)
Cheers,
Bijan Parsia.
Cheers,
Bijan Parsia.
Frank A. Adrian
> projects using different languages? Or do they study the ratio of how
> much projects cost, given the complexity of the application (in function
> points)?
> If some good studies indicated that Java had a significantly higher or
> even similar project success rate compared with Smalltalk (which I
> doubt), I would then have no problem going with Java.
The problem is that if you actually look at the figures, the language used
plays very little role in the success or failure of any given project. This
is not surprising, given that coding only contributes ~30% of the time to
most traditional (read semi-waterfall with highly spec'ed requirements)
projects. Bottom line, the main determinant of success is project length.
The longer the project's timeline, the less likely it is to be successful.
The reasons should be obvious - longer project = more money being spent =
more "features" needing to be included to make it politically palatable =
more risk of key team members leaving = ... = higher profile to be shot at
for political gain = <add your favorite dysfunction stemming from project
length here>. The language used to implement the project is a secondary
issue.
Now, that being said, this is also Smalltalk's advantage. Not only does it
have an ability to shorten that crucial 30% spent on implementation, but it
is also much easier to meld a highly iterative, quick turn around
methodology (read XP) to an interactive language like itself. The latter
feature can significantly reduce the risk of overall project failure. The
more Smalltalk ties itself to XP and shows that it is the best vehicle from
which to stage an XP project, the better off it will be.
I'm actually trying to show that a modified XP style using CLOS is even
better than a Smalltalk-based XP, but then, I'm a bit of an iconoclast when
it comes to programming matters, so I doubt the Smalltalk community has much
to fear about this.
faa
James A. Sawyer
> 2004.
What it doesn't say is that graduating "with Honors" means you were in a group
representing less than 10% of your class, and that being in a group representing
less than 5% of all intellectual activity would make you a Mensa candidate.
"5% of all OO activities" is absolutely meaningless without context.
Now if you had to select a context, which 5% do you suppose
the ST developed products will represent?
What does that tell you?
-jim
Anthony Woods
"...no new projects should start with Smalltalk..."
This is an interesting statement, GartnerGroup advised us in late 98
that Smalltalk was a Mature language with a good 10 years before it
reaches the 'end of life' stage. The document also states '...increase
tool usage'.
GG has also released in its 'Monthly Research Review' (dated 1st October
1998) that VA for Smalltalk is a Leader in the Smalltalk IDE market. It
further states: The advantages of using the VA product family include
working with a single, low-risk vendor that has a proven track record of
migrating its customers from older technologies into newer ones.....We
recommend that IBM customers develop migration strategies from
less-strategic tools to more-strategic ones, such as VA for Java and VA
Generator....
Being that GG sees VA Generator (a feature of VA Smalltalk) as a
strategic enterprise development area; I cannot see how one could remove
St from the equation.
I was unable to locate any information regarding the alleged demise of
Smalltalk on the GG website. By chance, do you have any information on
the official stance regarding St?
Anthony J Woods
In article <3888BA21...@acm.org>,
Thierry Thelliez <thi...@acm.org> wrote:
> Dear ST colleagues,
>
> Not a good week for us here. My management decided to get the advise
of
> the Gartner Group
> about IM tools. Of course Smalltalk (and GemStone/S) are not their
> prefered tools.
> GemStone/S was described as 'questionable' and no new projects should
> start with Smalltalk.
>
> Any idea on how to counter attack that ?
>
> Thierry
>
>
--
Anthony J Woods
Qantas Airways - The Spirit of Australia
l...@cc.gatech.edu
> Doug Way <dw...@mat.net> wrote in message news:86tjau$6n8$1...@nnrp1.deja.com...
> > Does the Gartner group have any studies which look at success rates for
> > projects using different languages? Or do they study the ratio of how
> > much projects cost, given the complexity of the application (in function
> > points)?
>
> > If some good studies indicated that Java had a significantly higher or
> > even similar project success rate compared with Smalltalk (which I
> > doubt), I would then have no problem going with Java.
>
> The problem is that if you actually look at the figures, the language used
> plays very little role in the success or failure of any given project. This
> is not surprising, given that coding only contributes ~30% of the time to
> most traditional (read semi-waterfall with highly spec'ed requirements)
> projects. Bottom line, the main determinant of success is project length.
> The longer the project's timeline, the less likely it is to be successful.
> The reasons should be obvious - longer project = more money being spent =
> more "features" needing to be included to make it politically palatable =
> more risk of key team members leaving = ... = higher profile to be shot at
> for political gain = <add your favorite dysfunction stemming from project
> length here>. The language used to implement the project is a secondary
> issue.
Thanks for the refreshing light. From this view, the reason that bad
languages live on is not purely from incompetence. It's just that
language really isn't the key factor in most situations.
Still, it would be nice to get the good languages being more popular.
I think what is really happening, is that people usually choose their
languagues when they are just learning to program--and long before
they have the sophistication to choose a good language. And heck,
more fundamentally, the bar for becoming a programmer is lower than
the bar for understanding how programming languages work. Or maybe
"different" rather than "lower". In any case, people commit to
languages early on, when they don't know better, and then it becomes
harder to switch later on. Who can afford to take six months off to
learn a new language, even if it sounds like a really nice one?
And then, you quite possibly have to go find a new place of work, if
you want to work in the new language! It's a real hassle leaving a
decent job to find another one, all because your current workplace has
standardized on some language you think is okay rather than great.
-Lex
Doug Way
bpa...@email.unc.edu (Bijan Parsia) wrote:
> Doug Way <dw...@mat.net> wrote:
>
> > I agree with your objection to this usage of "risk".
> [snip]
> > I would think the biggest risk involved would be project failure.
>
> Perhaps "most significant" risk.
That's what I meant. ("biggest" could be misinterpreted as "most likely
to occur", which isn't what I meant... oops.)
> So, for example, abstractly we can be reasonably sure that there are
> more Java programmers than Smalltalk programmers. But if I thereby
> decide to rule out Smalltalk, I might well miss that there is a *slew*
> of programmers *expert* in my problem domain who vastly prefer working
> in Smalltalk, and who would, perhaps, take a lower salary, or be worth
> more, etc.
>
> (Obviously, the reverse can be true. I'm just trying to illustrate
that
> these decisions *oughten* to be made on just wifty analysis, as the
> article suggested.)
>
> > *This* is the real
> > "risk" value. The smaller (but still significant) risk of using an
> > unpopular language would factor into this value, by virtue of the
> > difficulty involved with finding good programmers to complete the
> > project, etc.
>
> Right, taking into account the *specifics* of each situation.
You're right, the specifics are very important. (E.g., you probably
wouldn't consider Smalltalk for writing device drivers, C++ is bad for
writing multiplatform GUI business apps, etc.)
Of course, I don't know if the Gartner studies help at all with these
sorts of specifics...
> > Also, I heard a figure of Smalltalk having a 5-10% market share in
OO
> > development. Is that really so low? I certainly wouldn't be
concerned
> > with buying, say, a Chrysler car, even though their market share is
> > similar. Apple seems to be doing fine with this market share. You
> > could even argue that such a market share is self-sustaining (as
opposed
> > to dwindling), given that some people will always want some sort of
> > alternative to the mainstream, if they perceive it as superior.
When
> > you get under 1%, I would start to become concerned.
>
> Er...why? Market share alone doesn't determine market *size*. Would
you
> rather have 1% of a 10inch pie, or 25% of a 2inch pie :)
>
> Like popularity arguments, market share (which is intimately
connection
> with popularity, of course) arguments rely mostly on secondary
effects,
> fears of monopoly positions, etc.
>
> Which are real effects, of course, but the "oh you should program for
> Windows since you'll have a bigger market" tended to ignore the fact
> that for many products *you* would have a "bigger" market in the Mac
> world.
I dunno, I guess my main point was that a 5-10% market share isn't
really something to be worried about. I agree with your main points.
One other factor that I don't think gets considered very often is the
size of the city that your business is in. If you're in a smallish
city, language popularity becomes a bit more important, in the sense
that it is more difficult to hire people who know an "unpopular"
language. In a major metropolitan area, this isn't usually a big deal.
(E.g. if you're in New York City, it seems a bit ridiculous to worry
about Smalltalk "only" having a 5-10% market share.) Also, I guess city
size becomes less of a consideration as telecommuting becomes more
popular.