[Prev][Next][Index][Thread]

Re: Why I don't use Dylan



In article <7A5b5.89842$HK2.1646245@news20.bellglobal.com>, "Maury 
Markowitz" <maury_markowitz@hotmail.com> wrote:

> "Bruce Hoult" <bruce@hoult.org> wrote in message
> bruce-F44D2F.13163411072000@news.akl.ihug.co.nz">news:bruce-F44D2F.13163411072000@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



Follow-Ups: References: