Why I don't use Dylan
tej...@my-deja.com
in my spare time for something better. Dolphin has been improving a
lot over the past couple of years, and I like it more and more, so I'm
spending less and less time browsing for something better. Dylan
seemed very interesting, and I downloaded version 1 from Harlequin
sometime in the past. But I was disappointed at how slowly it compiled
and how big the dll's needed by the executable were. More recently, I
downloaded version 2 from F.O. and started to give it a try, by
following the examples and/or tutorials. I started at the beginning,
where it suggested starting with the Reversi example. But when I went
there, it said something to the effect that I had to pay money for
libraries before I could proceed with that. But since I'm just
browsing for something better than Smalltalk, I'm not ready to pay
money for it yet. Especially since Smalltalk keeps improving, and it
starts to seem like I will get the best by simply staying with
Smalltalk while it continues to improve.
And I think Dylan may gradually fade away, because it might not have
enough support to ever get critical mass. Critical mass is what makes
Java useful in spite of how bad a language it is. It causes the
vendors to compete against each other to provide the best tools at the
lowest prices, and the biggest and most functional libraries.
It's sad, because Dylan seemed almost like the ultimate programming
language, and I had high hopes for it. It's too bad the best
programming language designers tend to suck so badly at promoting their
products.
But even though I have lost a lot of my interest in Dylan, I still do
discuss it sometimes, and sometimes suggest to other programmers that
it might be worth trying, especially if they're willing to pay money to
try it. One thing I would like to know is how well it does in
benchmarks against LispWorks, because that question came up in a recent
discussion.
Sent via Deja.com http://www.deja.com/
Before you buy.
Dustin Voss
You can try all the libraries for, like, 30 days without paying a cent.
And critical mass is created by people using the product. Even Java
started from nothing. There's probably a good quote about that by
someone famous...
Timothy Guimond
I think it is a cool language. That said, I followed the same course as you
did.
I downloaded Harlequin Dylan 1 and when I saw Functional Objects I
downloaded the free version. I was very disappointed when I tried to
compile some of the examples, but I was still interested in seeing what
it was all about so I went back to the website and did a little more
reading.
I downloaded the Professional Combo Pack and contacte Functional
Objects for an unlock code. (You'll have some issues if you do download
and install the free version and then get one of the 30-day evals).
I installed the whole set of software and docs.
I must admit that it is a complete package. The price is reasonable.
The language seems complete and well thought out. It takes a little
getting used to because instead of providing many different ways
to get something done, Dylan gives you one process that is used to
do many different things. The libraries seem to have alot of functionality
included. Object orientated programming is done a little different but it
kind of grows on you after a while.
It won't replace Smalltalk in the near future, but it is certainly worth
a look. Free evaluations allow you to try before you buy and if you
don't like it, remove it and move on.
Give it another look.
Tim
<tej...@my-deja.com> wrote in message news:8je1hp$53j$1...@nnrp1.deja.com...
oo...@my-deja.com
30 days. Installing is pain-free... unlike with some other free
evaluations. They don't even really make you register.
Functional Developer has improved a lot since the Harlequin Dylan 1.0
days. Give it a try.
Don't forget to install the Microsoft linker which you can download from
an SDK at Microsoft's web site. It speeds up the building process by
about twice because the GNU linker is so slow.
Anyway, the cost of Functional Developer for many will be trivial
compared to the productivity gain.
Scott McKay
I also can't fathom for the life of me why someone would simultaneously
think that Dylan might almost be the ultimate programming language,
but then won't even take the trouble to see that there is a free evaluation,
and furthermore spreads misinformation to other people about them
having to pay money even to try it.
I don't work for Functional Objects, so don't hold my sharp tongue against
them.
tej...@my-deja.com wrote in message <8je1hp$53j$1...@nnrp1.deja.com>...
tej...@my-deja.com
"Timothy Guimond" <tgui...@ma.ultranet.com> wrote:
> a look. Free evaluations allow you to try before you buy and if you
> don't like it, remove it and move on.
>
> Give it another look.
The problem is the 30-day limit. I only have a small amount of spare
time per week. I have to wait till sometime in the future when I have
30 days of free time.
I also want to evalutate other languages, such as OCaml. I will
probably start that soon, but have no idea how long I will continue
with it. I have a hunch that I will probably end up still using
Smalltalk for the foreseeable future, because it keeps improving and
becoming more popular.
When ParcPlace/ObjectShare had financial problems, they sold their main
Smalltalk product, VisualWorks, for less than $1 million. If the same
thing happens to Functional Objects, I doubt they could even get that
much, because VisualWorks was a proven product with a well-established
market. What if some company such as Sun buys F.D. for a bargain
price, and uses it for internal research only, like they did with Self?
I've already invested a lot of time into Smalltalk, which seems like a
good investment. But common sense tells me Dylan is not the same kind
of investment at all. I do want to get involved with Dylan, when I
have time, but I don't think I should delude myself into thinking it
will be anything but a hobby for me. Meanwhile Smalltalk is how I earn
my living, and I have to give it a lot more respect and attention than
it probably deserves, at the expense of other possibly more interesting
languages such as OCaml and Dylan.
Charles Hixson
> In article <8jehcr$5ul$1...@bob.news.rcn.net>,
> "Timothy Guimond" <tgui...@ma.ultranet.com> wrote:
> > a look. Free evaluations allow you to try before you buy and if you
> > don't like it, remove it and move on.
> >
> > Give it another look.
>
> The problem is the 30-day limit. I only have a small amount of spare
> time per week. I have to wait till sometime in the future when I have
> 30 days of free time.
Second! Demo versions should be limited by the number of uses, not the
number of days. I keep never seeing that I will have enough free time in
the next period to do a proper evaluation, so I never get around to
installing the downloads.
The last time I bought Dylan I eventually gave up on the macro syntax. I
never did get the dialog to display, though I followed the example
exactly. Presumably things have been fixed since then, but now I'm mainly
looking at other languages (and Harlequin has "spun off" it's Dylan
compiler). Python, Ruby, Ada95 ... If I could blend Ruby and Ada95....
Dylan remains very interesting, but FO Dylan is limited to Windows, and I
would prefer to move to Linux. I don't know how compatible FO Dylan is to
gwydion (or what state gwydion is currently in). Etc.
Limited time. Lots of choices. A lot of what I need to do involves screen
design, databases, and printing reports. If it weren't for MS and VB, then
Access would be the obvious choice, but Basic is still a horrible language,
and MS changes the file formats incompatibly between every version. And
the errors! So I really want out of the entire MS system. (Not to mention
UCITA, and what *THAT* implies, or that MS is one of its big supporters,
and what THAT implies.)
Rob Myers
>
> tej...@my-deja.com wrote:
>
> > The problem is the 30-day limit. I only have a small amount of spare
> > time per week. I have to wait till sometime in the future when I have
> > 30 days of free time.
>
> Second! Demo versions should be limited by the number of uses, not the
> number of days.
It is a common practice in the software industry, though. If you don't
like it, email Fun-O and tell them how a different system would
encourage you to evaluate their products.
> The last time I bought Dylan I eventually gave up on the macro syntax.
I can relate to that. :-) It looks so ugly and unlike the rest of Dylan
at first.
If you look through the Dylan library open sources at gwydiondylan.org,
the examples can help you make the leap much more quickly (I'd have been
lost without them). I can give you the urls if you're interested.
> Limited time. Lots of choices. A lot of what I need to do involves screen
> design, databases, and printing reports.
You can do this in Functional Developer, I think. There was a
testimonial from a database report project done in Functional Developer
on this newsgroup a few months back.
> If it weren't for MS and VB, then Access would be the obvious choice
Functional developer has ODBC, its own object database, the ability to
call Win32 APIs, its own GUI system and lots of other features it sounds
like you are after. I don't know enough about printing or screen
painting to comment on these, though.
- Rob.
--
Rob Myers - http://www.robmyers.org/ H2G2 - http://www.h2g2.com/
MacOS wonderfulness for The Hitch Hiker's Guide to the Galaxy Game.
"Don't talk to sociologists. Social practice has no sociological
content." - Art & Language.
Peter Westlake
s...@mediaone.net says...
> So there are 30-day free evaluations for everything from Fun-O.
>
> I also can't fathom for the life of me why someone would simultaneously
> think that Dylan might almost be the ultimate programming language,
> but then won't even take the trouble to see that there is a free evaluation,
> and furthermore spreads misinformation to other people about them
> having to pay money even to try it.
> >downloaded version 2 from F.O. and started to give it a try, by
> >following the examples and/or tutorials. I started at the beginning,
> >where it suggested starting with the Reversi example. But when I went
> >there, it said something to the effect that I had to pay money for
> >libraries before I could proceed with that
Okay, so tejavu missed that the libraries could be tried free. But apart
from that, I think the point is valid, because that same thing tripped me
up too. The point is that this is *the main example* at the *beginning*
of the help for the free product. If I download a free C compiler, I
don't expect to have to download a separate, time-limited trial version
of some other thing in order to try "hello, world", especially if the
help hasn't mentioned it. And yes, I know there's a "hello world" program
included, but it isn't what the Quick Start topic in the help files uses.
Another gotcha, while we're on the subject, is the way the example
programs are paused so that you can read their output. If the dialog box
that tells you they are paused was to say "and you can un-pause them by
clicking the "start" button again", that would be an improvement in
usability out of all proportion to the effort involved.
Peter.
--
peter@harle Harlequin Ltd, Barrington Hall,
quin.co.uk Barrington, Cambs CB2 5RG, England.
tej...@my-deja.com
> I must admit that it is a complete package. The price is reasonable.
But it's hard for most programmers to take it seriously. Functional
Objects does not inspire confidence that the programmer would be making
a wise investment of time and money. The company is run by former
Harlequin employees who give the vague impression they might be doing
it as a part time hobby. There have been so many such products in the
past which have gradually faded into obscurity, when their proprietors
found their time occupied by other pursuits and put the product on the
back burner, till they gradually reached the point where they could no
longer even do any more maintenance on it.
One thing Functional Objects could do to improve their credibility a
lot would be to put their source code in escrow with SourceForge to
become free open source if Functional Objects ever drops the ball. A
clear guarantee, that the product will definitely have a future no
matter what, is exactly what people need to be able to take it
seriously enough to invest the time and money needed to get started
with it.
What is the development situation at Functional Objects? I get the
vague impression some contract programmers working for other companies
might be doing the Functional Developer work between contracts and/or
in their spare time. Or does Functional Objects have venture capital
to pay the salaries of full time programmers?
Rob Myers
> Subject: Re: Why I don't use Dylan
>
> "Timothy Guimond" <tgui...@ma.ultranet.com> wrote:
>
>> I must admit that it is a complete package. The price is reasonable.
>
> But it's hard for most programmers to take it seriously. Functional
> Objects does not inspire confidence that the programmer would be making
> a wise investment of time and money. The company is run by former
> Harlequin employees who give the vague impression they might be doing
> it as a part time hobby.
Having communicated with Functional Objects personnel in both public and
private email, and having read through the business plans and white papers
on their web site, I can assure you that I have not got this impression.
They seem to be a hard working group of very capable people dedicated to
moving Functional Developer forward.
Ex-Harlequin personnel not with Functional Objects do answer questions on
this list as far as I can tell. I'd take that as a vote of confidence rather
than anything else.
> There have been so many such products in the
> past which have gradually faded into obscurity, when their proprietors
> found their time occupied by other pursuits and put the product on the
> back burner, till they gradually reached the point where they could no
> longer even do any more maintenance on it.
I can't see evidence of that happening here. Fun-O have already finalized
and released the 2.0 version of Functional Developer and have announced
plans for a UNIX version.
> One thing Functional Objects could do to improve their credibility a
> lot would be to put their source code in escrow with SourceForge to
> become free open source if Functional Objects ever drops the ball. A
> clear guarantee, that the product will definitely have a future no
> matter what, is exactly what people need to be able to take it
> seriously enough to invest the time and money needed to get started
> with it.
SourceForge has an air of "amateur project" about it that I'm not sure would
help. The public sources are freely available with the Functional Developer
downloads, so if a neutron bomb does get dropped near their offices, you'll
still have them. Some are even on the Gwydion site and CVS server.
> What is the development situation at Functional Objects? I get the
> vague impression some contract programmers working for other companies
> might be doing the Functional Developer work between contracts and/or
> in their spare time. Or does Functional Objects have venture capital
> to pay the salaries of full time programmers?
This is company confidential stuff for a start-up, so I can't imagine them
letting us know.
I was worried when Harlequin got split up, but Functional Objects have come
through with flying colours. My only small concern is that their website
could do with making more boring-corporate, but it has grown on me :-)
- Rob.
Rainer Joswig
<robm...@cwcom.net> wrote:
> I was worried when Harlequin got split up, but Functional Objects have come
> through with flying colours. My only small concern is that their website
> could do with making more boring-corporate, but it has grown on me :-)
I'm neither a customer nor am I known to be very sympathetic
to Dylan in general. But I really like how the product is being
sold in pieces by "Functional Objects, Inc." over their web site.
This really makes sense to me in the Internet age.
Make libraries affordable and let customers buy libraries
based on their real demand - this is really a good thing.
--
Rainer Joswig, BU Partner,
ISION Internet AG, Steinhöft 9, 20459 Hamburg, Germany
Tel: +49 40 3070 2950, Fax: +49 40 3070 2999
Email: mailto:rainer...@ision.net WWW: http://www.ision.net/
Maury Markowitz
news:B58F68FF.4091%robm...@cwcom.net...
> Having communicated with Functional Objects personnel in both public and
> private email, and having read through the business plans and white papers
> on their web site, I can assure you that I have not got this impression.
> They seem to be a hard working group of very capable people dedicated to
> moving Functional Developer forward.
I hope you don't mind me jumping in here.
For some reason that I don't understand, computer programming is based on
a popularity contest. The faintest whiff of suspicion on the future of any
development system is enough to send it into the trash by the vast majority
of the user it's targetted at. Sometimes this is deserved, consider
ProGraph's almost-comical series of "we're back!" followed by "sorry, dead
again".
However, consider Dylan at Apple. Seems to me that the vast majority of
Mac users are still better off on Dylan than they are on, say, C++ for the
"average" project, and yet it's usage is effectively zero. There's a number
of good reasons for this, but an equal number of bad ones that are likely
more often the case.
My point is that I'm not sure Ex-Harlequin personnel mean anything one way
or the other. They do not have the market clout to have a serious name, and
as a result I'd suspect they'll shortly have "day jobs" like Pictorius does
today, if not already. And that just makes the matter worse.
Too bad really, if only BillJ wanted a new Dylan rather than a new C++...
> SourceForge has an air of "amateur project" about it that I'm not sure
would
> help. The public sources are freely available with the Functional
Developer
> downloads, so if a neutron bomb does get dropped near their offices,
you'll
> still have them. Some are even on the Gwydion site and CVS server.
Methinks you doth protest too much. Let me assure you that escrow on
SourceForce would help credibility far more than it would hurt! This is an
excellent suggestion, you shouldn't write it off so easily.
> This is company confidential stuff for a start-up, so I can't imagine them
> letting us know.
Sounds to me like you're all too willing to be their appologist. The
issue here is credibility, I'd say that the company would be happy to tell
you if you asked in order to support just that.
Maury
Bruce Hoult
Markowitz" <maury_m...@hotmail.com> wrote:
> For some reason that I don't understand, computer programming
> is based on a popularity contest. The faintest whiff of suspicion
> on the future of any development system is enough to send it into
> the trash by the vast majority of the user it's targetted at.
> Sometimes this is deserved, consider ProGraph's almost-comical
> series of "we're back!" followed by "sorry, dead again".
Hard to avoid, I think. It takes a large investment of time to learn a
new programming language, and once you build up your own library of
useful stuff you really don't want to find that the company that
produced your programming language is dead and that a new OS version is
incompatable with what you're using, or that show-stopping bug will
*never* be fixed.
It's amazing how long THINK Pascal kept on keeping on years and years
after it was no longer developed, sold or supported, but something in
about MacOS 8.1 or so finally killed it and it's dead, dead, dead today,
even thought it's *still* probably the best ever Pascal or C learning
environment.
Which is why I'm so excited about Gwydion Dylan. Right now it's less
cosmetically polished than either the Apple Dylan or Functional
Developer products, but it's also very bug-free and produces good code
and produces better code with every month that goes by. And because I
have the source code and it's a simple command-line application I can
make it work on any system that I want to use in the future, and can fix
any bugs myself.
> However, consider Dylan at Apple. Seems to me that the vast
> majority of Mac users are still better off on Dylan than they
> are on, say, C++ for the "average" project,
I certainly think they would be.
> and yet it's usage is effectively zero. There's a number
> of good reasons for this, but an equal number of bad ones that are likely
> more often the case.
Where would we be now if Apple has included the source code on the Dylan
Technology Release CD?
-- Bruce
Rob Myers
>
> "Rob Myers" <robm...@cwcom.net> wrote in message
> news:B58F68FF.4091%robm...@cwcom.net...
>
> > would help.
>
> SourceForce would help credibility far more than it would hurt! This is an
> excellent suggestion, you shouldn't write it off so easily.
OK, I'll take a look at SourceForge again. I'm possibly stuck in the
early web-era snobbery of everything having to have its own domain name.
What happens if SourceForge's business model is broken and they go down? :-)
>
> > This is company confidential stuff for a start-up, so I can't imagine them
> > letting us know.
>
> Sounds to me like you're all too willing to be their appologist.
Eep! I wouldn't want to be an apologist for anything! The background to
my comments here is that I've been involved in a couple of startups now
and you tend to keep your cards prety close to your chest unless there's
a very good reason not to. Responding to concerns about credibility can
harm credibility.:-)
Maury Markowitz
news:bruce-F44D2F....@news.akl.ihug.co.nz...
> Hard to avoid, I think. It takes a large investment of time to learn a
> new programming language
Well I don't know about that. I'd say that a change in language is smaller
in effort to learn than a change in API. I picked up Obj-C in a couple of
days. Sure it's different when you have some monstrosity like C++ or Ada,
but with a well designed language and libs, I'm not at all convinced that
this is a big issue.
The real problem, IMHO, is that everyone is burned out by C++. Now all
OOPS programming will forever be tainted by the effort put into learning C++
over the last 5 years or so. But in the 80's it was "fun" to pick up new
languages.
> produced your programming language is dead and that a new OS version is
> incompatable with what you're using, or that show-stopping bug will
> *never* be fixed.
Sure, which is why I think open source is a requirement for all
development efforts of both OS's and the dev systems that serve them. Why
oh WHY can't AT&T drop the pickle on Plan 9 after all these years?!
> It's amazing how long THINK Pascal kept on keeping on years and years
> after it was no longer developed, sold or supported, but something in
> about MacOS 8.1 or so finally killed it and it's dead, dead, dead today,
> even thought it's *still* probably the best ever Pascal or C learning
> environment.
Yeah, Lightspeed was cool too.
> Where would we be now if Apple has included the source code on the Dylan
> Technology Release CD?
Hmmm, interesting question. My only concern here is that, again IMHO, I
see a lot of herdism in the MacOS community. While everyone makes fun of
Apple's NIH, their developers were just as bad if not worse (myself
included, but I lost my religion in 98 and I'm better off for it). Consider
the outright shunning of things like MagicCap and WebTV, both in-Apple
ideas. Wild. Well, there's been a lot of culling there anyway.
I think Cocoa serves the Apple world very well, although I think it would
be even better with a Win version. However I do have problems with the
Obj-C language that I would like to see addressed. So OK, what would it
take to make Dylan compile onto a Obj-C like runtime? Is this a
possibility? If so, we all need to talk!
Maury
Bruce Hoult
Markowitz" <maury_m...@hotmail.com> wrote:
> "Bruce Hoult" <br...@hoult.org> wrote in message
> news:bruce-F44D2F....@news.akl.ihug.co.nz...
> > Hard to avoid, I think. It takes a large investment of time to learn a
> > new programming language
>
> Well I don't know about that. I'd say that a change in language is
> smaller in effort to learn than a change in API.
Except when they go together e.g. Java, Cocoa, SmallTalk.
> I picked up Obj-C in a couple of days. Sure it's different when
> you have some monstrosity like C++ or Ada, but with a well
> designed language and libs, I'm not at all convinced that this
> is a big issue.
Probably true. Which well-designed language and libs that has acheived
immense populatiry in the last decade do you have in mind?
> The real problem, IMHO, is that everyone is burned out by C++. Now all
> OOPS programming will forever be tainted by the effort put into learning
> C++ over the last 5 years or so.
5? I seem to recall starting to use CFront with MacApp in 1990 or so.
> I think Cocoa serves the Apple world very well, although I think it
> would be even better with a Win version. However I do have problems
> with the Obj-C language that I would like to see addressed. So OK,
> what would it take to make Dylan compile onto a Obj-C like runtime?
> Is this a possibility? If so, we all need to talk!
I really must take a look at MacOS X soon. I'm still doing a lot of Mac
development (e.g. my current contract is for a Mac application), but I'm
using CW Pro 4, MacOS 8.6, and haven't been a paying member of Apple's
developer program for a couple of years. Maybe I'm just still
shell-shocked by learning and then abandoning too many technologies
(OpenDoc, Copland, you know the list ...) in the past. Shouldn't be too
hard as I've done a lot of Solaris and Linux work in the last several
years :-)
Anyway, the gist is that I don't know what the Obj-C runtime is like.
My understanding from early NeXT days was that the compiler just gathers
up those Smalltalk-style method selector names into a string, looks them
up in a table to find a numeric selector ID, and adds that as the first
argument to a call to a global "dispatch" function, using a standard C
function call.
Is that correct? Has it changed in the last ten years?
-- Bruce
Rob Myers
> Anyway, the gist is that I don't know what the Obj-C runtime is like.
> My understanding from early NeXT days was that the compiler just gathers
> up those Smalltalk-style method selector names into a string, looks them
> up in a table to find a numeric selector ID, and adds that as the first
> argument to a call to a global "dispatch" function, using a standard C
> function call.
AFAIK, there's some speed-improving cacheing of id's, but the C calls
are still there. This should make it easy to glue d2c to Objective-C,
and indeed allow anything that can call C code to call Objective-C. The
same can't be said for C++ (name mangling) or Java (name manging and
type mapping).
So calling Cocoa or OpenStep code from Dylan should be achievable
without too much hacking.
- Rob.
--
Rob Myers - http://www.robmyers.org/ H2G2 - http://www.h2g2.com/
MacOS wonderfulness for The Hitch Hiker's Guide to the Galaxy Game.
"Smash global capitalism! Spend less money!"
Maury Markowitz
> > Well I don't know about that. I'd say that a change in language is
> > smaller in effort to learn than a change in API.
>
> Except when they go together e.g. Java, Cocoa, SmallTalk.
No way Bruce, you should come to one of my talks some time.
I had _zero_ "real" programming experience prior to learning Cocoa, I
knew HyperCard and stuff like that. I had been trying to break into "real"
programming for years, trying product after product in hopes of lowering the
bar to the point where I could get involved without a change of lifestyle.
I tried it all, yes, even Apple Dylan, but everything fell down at one point
or another and I was no closer than before.
Then I tried Cocoa. Five days later I was working on commercial code, and
shipped our product six months later.
Cocoa, due to being self-contained, is _rediculously_ easy to learn. I
disagree completely with your point above.
> Probably true. Which well-designed language and libs that has acheived
> immense populatiry in the last decade do you have in mind?
None. I'm not sure what your point is here though. As I said, this is a
popularity contest, the best tech rarely wins.
> 5? I seem to recall starting to use CFront with MacApp in 1990 or so.
Sure, and I was playing with MacApp 3 way back too. However the market as
a whole was largely pure C until more recently than that.
> Anyway, the gist is that I don't know what the Obj-C runtime is like.
Several things...
1) method calls based pointers to a class, then a hash lookup - the hash can
be changed at runtime
1a) method signatures are hashed off the method name and parameter _names_
(note!) and thus method overloading is not supported in the current runtime
(big problem IMHO).
2) class lookup via a similar hash, but only once and then cached in the
instance
3) "two try" dispatch, if the method isn't supported, try the method
"perform" instead, which everyone supports - zero effort forwarding and
other nifty features come from this
4) dynamic loading/unloading of basically everything.
Missing...
1) GC
2) namespace support
3) ivars are not dynamic
4) no support for MI (likely a no-brainer with (2) though)
Maury
Maury Markowitz
news:396D91D5...@h2g2.com...
> So calling Cocoa or OpenStep code from Dylan should be achievable
> without too much hacking.
Hmmm, this would certainly be interesting.
Maury