C# should be last plain-text language

C# should be last plain-text language

[Amir Yantimirov]

http://www174.pair.com/yamir/programming/sharplast.htm

Actually this article is not about C#.
It's my view of how program development should be organized on any language.
Some of proposed features I didn't see in any IDE including VS.Net, 
Borland Builder, screenshots of KDevelop and GPS.
Also, "Standard ADA Preprocessor" topic somewhere relevant.

Amir Yantimirov
http://www174.pair.com/yamir/programming

Re: C# should be last plain-text language

[Peter Amey]

Amir Yantimirov wrote:
> http://www174.pair.com/yamir/programming/sharplast.htm
> 
> Actually this article is not about C#.
> It's my view of how program development should be organized on any language.
> Some of proposed features I didn't see in any IDE including VS.Net, 
> Borland Builder, screenshots of KDevelop and GPS.
> Also, "Standard ADA Preprocessor" topic somewhere relevant.
> 
> Amir Yantimirov
> http://www174.pair.com/yamir/programming

Looks rather like "literate programming" [Knuth 1984].

Peter

Re: C# should be last plain-text language

[Marius Amado Alves]

Peter Amey wrote:
> Amir Yantimirov wrote:
> > http://www174.pair.com/yamir/programming/sharplast.htm
> > It's my view of how program development should be organized on any
> > language....
> Looks rather like "literate programming" [Knuth 1984].

To me looks more like a specification document for something like jGRASP.

Re: C# should be last plain-text language

[Larry Barowski]

> "Marius Amado Alves" <maa@liacc.up.pt> wrote in message
news:mailman.1.1076068084.295.comp.lang.ada@ada-france.org...
> Peter Amey wrote:
> > Amir Yantimirov wrote:
> > > http://www174.pair.com/yamir/programming/sharplast.htm
> > > It's my view of how program development should be organized on any
> > > language....
> > Looks rather like "literate programming" [Knuth 1984].
>
> To me looks more like a specification document for something like jGRASP.

From the web page, "XML advocates can propose some format to

extend existent programming languages (XADA, XC++, XC# ...)

and express generation of any view or pure files through

transformations. Though I personally prefer binary formats

I think XML solution have more chanses to be accepted by

broad community. The other possible way is including some

META-comments in source files."



It seems to me that you don't need any of those, and would be

better off doing this another way. All the views and editing

features you want can be created on demand from plain text

source code, with only a "special comment" syntax, like

javadoc comments, for structure-specific comments and one

for "ability to mark declaration as commented out (for debug

purpose)". You can store information on the current state of

the views (what is folded, etc.) in a separate database. If

someone edits the code outside of your tool, they will lose that

state (no big deal), unless the other editor is compatible with

your tool. This way developers will be able to work with raw

source code and other editing tools if they want to (and they will

want to), but still have the views and features you want. From

the point of view of someone who uses your tool exclusively,

there is essentially no difference between a META-language or

META-comments and using a database. This is the method we

use to store source code folding information, breakpoint locations,

etc. in jGRASP.



-Larry Barowski

Re: C# should be last plain-text language

[Larry Barowski]

Sorry for the formatting above - I need a better news reader.

Re: C# should be last plain-text language

[Amir Yantimirov]

"Larry Barowski" <larrybarATengDOTauburnDOTeduANDthatISall> wrote in message news:<1028f364e6smfca@corp.supernews.com>...

> It seems to me that you don't need any of those, and would be
> better off doing this another way. All the views and editing
> features you want can be created on demand from plain text
> source code, with only a "special comment" syntax, like
> javadoc comments, for structure-specific comments and one
> for "ability to mark declaration as commented out (for debug
> purpose)". You can store information on the current state of
> the views (what is folded, etc.) in a separate database. If
> someone edits the code outside of your tool, they will lose that
> state (no big deal), unless the other editor is compatible with
> your tool. This way developers will be able to work with raw
> source code and other editing tools if they want to (and they will
> want to), but still have the views and features you want. From
> the point of view of someone who uses your tool exclusively,
> there is essentially no difference between a META-language or
> META-comments and using a database. This is the method we
> use to store source code folding information, breakpoint locations,
> etc. in jGRASP.
> 
> -Larry Barowski

You are right, of course. But my intention is what developer 
should not work with raw code at all. The raw code should be 
something alike intermediate assembler code in some compilers.
And then separated database dont't have much benefits.

Generally, I think, existent programming languages are good enough
for solving current programing tasks. Troubles come not from what tasks
are difficult but from what they are numerous. This problem lies 
outside of language domain. And any new improvements as Ada 20XX or
C++ 20XX cannot change this fact.

The lesson of most productive environment I work with (C++ Builder)
is that language is only part of development system. And the most 
effective addition to language are those what helps language better  
integrate with whole system. The cost in case is nonportatibility.

By the way, I looked at http://www.jgrasp.org/
but from PDFs I see the same alphabetical list of files I argue against. :)

Amir Yantimirov
http://www174.pair.com/yamir/programming/

Re: C# should be last plain-text language

[Larry Barowski]

>
> You are right, of course. But my intention is what developer
> should not work with raw code at all. The raw code should be
> something alike intermediate assembler code in some compilers.
> And then separated database dont't have much benefits.
>

It has the huge benefit of compatibility with existing tools
and systems without the need for "importing and exporting"
of raw source or cluttering of source with meta-comments.
If no one uses your tool, it will be useless in promoting the
programming methodology it supports. Those using the tool
exclusively would not have to work with raw source code.

> By the way, I looked at http://www.jgrasp.org/
> but from PDFs I see the same alphabetical list of files I argue against.
:)
>  ...

Well, the focus of jGRASP is low and high level
visualizations. Viewing or editing source code without
having a frequently updated control flow graph in the
indentation space is something we argue against.

-Larry Barowski

Re: C# should be last plain-text language

[Alexandre E. Kopilovitch]

Amir Yantimirov wrote:

> Generally, I think, existent programming languages are good enough
> for solving current programing tasks. Troubles come not from what tasks
> are difficult but from what they are numerous. This problem lies 
> outside of language domain. And any new improvements as Ada 20XX or
> C++ 20XX cannot change this fact.
>
> The lesson of most productive environment I work with (C++ Builder)
> is that language is only part of development system. And the most 
> effective addition to language are those what helps language better  
> integrate with whole system.

As you mentioned (Borland) C++ Builder, you must know the language element,
which is fairly critical for Delphi and then C++ Builder integration abilities
- "property". I always thought that this language element would be useful in
Ada, but it does not seem a simple task to add it properly. I tried that
several times, but failed so far. 

	 


Alexander Kopilovitch                      aek@vib.usr.pu.ru
Saint-Petersburg
Russia

Re: C# should be last plain-text language

[Amir Yantimirov]

"Alexandre E. Kopilovitch" <aek@VB1162.spb.edu> wrote in message news:<mailman.5.1076254282.295.comp.lang.ada@ada-france.org>...
> As you mentioned (Borland) C++ Builder, you must know the language element,
> which is fairly critical for Delphi and then C++ Builder integration abilities
> - "property". I always thought that this language element would be useful in
> Ada, but it does not seem a simple task to add it properly. I tried that
> several times, but failed so far. 
>
> Alexander Kopilovitch                      aek@vib.usr.pu.ru
> Saint-Petersburg
> Russia

The property is key element of design-time interface of object in 
Delphi/Builder/VS.Net. At run-time it may be convenient but not necessary.
Developers of A# (http://www.usafa.af.mil/dfcs/bios/mcc_html/a_sharp.html)
use same approach as Java - get_Value/set_Value naming convention.

Amir Yantimirov
http://www174.pair.com/yamir/programming/