www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Syntax question...

reply "RA" <ra wolven.net> writes:
	charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

I was curious about how hard it would be to create a different =
=93language
front end=94 for D? =20
Since you (Walter) have made the compiler front end (lexer, parser, etc)
source code available,
would it be possible for someone to modify it and create a different =
(for
example, something=20
more like VB) language syntax that would still work with the back end
compiler?
=20
I realize most of you long time C\C++ programmers probably couldn=92t =
see any
need for=20
such a thing, but others of us (at least me) would prefer something not
quite as cryptic=20
as C like syntax.
=20
I=92ve never written a compiler\lexer\parser\etc. so I don=92t know what =
all
would be involved=20
in creating a different syntax.  In theory, it doesn=92t seem like it =
should
be =93all=94 that hard=85
But then, naivet=E9 is great isn=92t it.  :-)
Oct 30 2006
next sibling parent reply "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
"RA" <ra wolven.net> wrote in message 
news:mailman.2.1162241354.24029.digitalmars-d puremagic.com...

---------------------------
I was curious about how hard it would be to create a different "language 
front end" for D?
Since you (Walter) have made the compiler front end (lexer, parser, etc) 
source code available,
would it be possible for someone to modify it and create a different (for 
example, something
more like VB) language syntax that would still work with the back end 
compiler?

I realize most of you long time C\C++ programmers probably couldn't see any 
need for
such a thing, but others of us (at least me) would prefer something not 
quite as cryptic
as C like syntax.

I've never written a compiler\lexer\parser\etc. so I don't know what all 
would be involved
in creating a different syntax.  In theory, it doesn't seem like it should 
be "all" that hard.
But then, naiveté is great isn't it.  J
---------------------------

(as a side note, it's kind of .. etiquette here to use plain text messages. 
some newsreaders don't take kindly to the HTML messages.)

It shouldn't be too incredibly difficult.  Probably the main problem would 
be coming up with some new syntaxes for things which commonly don't exist in 
Basic languages, or for which there's no clear standard.  For example, in a 
lot of Basic languages I've used, arrays are indexed with parentheses, but 
that would mean that you wouldn't be able to have opCall and 
opIndex/IndexAssign in the same class, so we'd have to use the C-style 
square brackets for that.

Another big problem is that some things that look cryptic in C-style syntax 
would look just as cryptic in Basic-style syntax, but with words instead of 
symbols.  So you wouldn't really be gaining that much.

Although, when it really comes down to it, Basic and C syntaxes aren't too 
different.  You've got basically all the same control structures, just with 
"control .. endcontrol" rather than "control { .. }".

Lastly there's the issue of case sensitivity.  I know a lot of Basic 
languages are not case sensitive, and so foo, Foo, and FOO are all the same 
thing.  But they're all different in D.  The same goes for the control 
structures -- if we _did_ make this DBasic dialect case sensitive, would we 
make the keywords and control structures all lowercase, camelcase, or all 
caps (AAAAAAGH!)?

So.. let's see what we've got.  Assuming that we decide that the DBasic 
dialect would be case-sensitive, with camelcase keywords..

Import std.stdio

Function main(args As Char[][]) As Void
    writefln("Hello World, Reloaded")

    'auto type inference and built-in foreach
    For Each(argc, argv; args)
        'Object Oriented Programming
        Dim cl As CmdLin = New CmdLin(argc, argv)

        'Improved typesafe printf
        writefln(cl.argnum, cl.suffix, " arg: %s", cl.argv)

        'Automatic or explicit memory management
        Delete cl
    Next Each

    'Nested structs and classes
    Struct specs
        'all members automatically initialized
        Field count As Int
        Field allocated As Int
    End Struct

    'Nested functions can refer to outer
    'variables like args
    Function argspecs() As Specs
        Local s as Specs* = New specs

        'no need for '->'
        s.count = args.length              'get length of array with .length
        s.allocated = TypeOf(args).sizeof  'built-in native type properties

        For Each(argv; args)
            s.allocated += argv.length * TypeOf(argv[0]).sizeof
        Next Each

        Return *s
    End Function

    'built-in string and common string operations
    writefln("argc = %d, " ~ "allocated = %d", argspecs().count, 
argspecs().allocated);
End Function

Class CmdLin
    Private Field _argc As Int
    Private Field _argv As Char[]

Public:
    'constructor
    Me(argc As Int, argv As Char[])
        _argc = argc
        _argv = argv
    End Me

    Function argnum() As Int
        return _argc + 1
    End Function

    Function argv() As Char[]
        return _argv
    End Function

    Function suffix() As Char[]
        Local suffix As Char[] = "th"

        Select(_argc)
            Case 0:
                suffix = "st"
                Break

            Case 1:
                suffix = "nd"
                Break

            Case 2:
                suffix = "rd"
                Break

            Default:
                Break
        End Select

        Return suffix
    End Function
End Class


Though when you're dealing with something like

Foo!(delegate int(int[char[]] arr) { return arr["hi"]; });

That would end up as..

'For lack of a better template instantiation syntax..
Foo!(Delegate(arr As Int[Char[]]) As Int Return arr["hi"] End Delegate)

That's a little cryptic too.. 
Oct 30 2006
parent reply %u <ra wolven.net> writes:
Thanx for the reply Jarret, I was beginning to feel a little ignored there... 
And
sorry bout the html stuff.  I didn't realize there would be a problem until I
saw
my post on here, then I changed my email settings.  I set them to rich text. Can
this news group handle that or should I just set it to plain text?

You gave some great samples there of VB style code, but I was just using VB as
an
example... it's not really what I'm wanting to do.  Basically, I want a "new"
syntax.  Although it would borrow heavily from other languages that make sense
to me.

So what exactly does not "too incredibly difficult" mean?  Which would be
easier,
modifying the existing front end source code, or doing as Hasan suggested and
writing a front end for the D front end...    I don't really like that concept,
but maybe it would be easier?
Oct 30 2006
parent reply "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
"%u" <ra wolven.net> wrote in message 
news:ei68it$2e9a$1 digitaldaemon.com...
 Thanx for the reply Jarret, I was beginning to feel a little ignored 
 there...  And
 sorry bout the html stuff.  I didn't realize there would be a problem 
 until I saw
 my post on here, then I changed my email settings.  I set them to rich 
 text. Can
 this news group handle that or should I just set it to plain text?
If this message that I'm replying to is rich text, then it looks through my reader and on the web interface. Have a look at your original post through the web interface (http://www.digitalmars.com/pnews/read.php?server=news.digitalmars.com&group=digital ars.D&artnum=43421) and you'll see why HTML posting is generally discouraged :S
 You gave some great samples there of VB style code, but I was just using 
 VB as an
 example... it's not really what I'm wanting to do.  Basically, I want a 
 "new"
 syntax.  Although it would borrow heavily from other languages that make 
 sense to me.
What kind of syntax are you thinking? (I'm genuinely interested :) )
 So what exactly does not "too incredibly difficult" mean?  Which would be 
 easier,
 modifying the existing front end source code, or doing as Hasan suggested 
 and
 writing a front end for the D front end...    I don't really like that 
 concept,
 but maybe it would be easier?
Well the front end source is... kind of messy, to say the least. Thankfully about all you'd have to mess with are the lexer and parser bits (lex.c and parse.c, imagine that), which aren't _too_ bad. Once you've got yourself a nice Basic-style lexer/parser set up, you could compile GDC with it and test it out. Of course, that might be a bit daunting, so.. What Hasan suggested is certainly not a bad idea either, and would also mean you wouldn't have to recompile the D compiler in order to get it to work. However, it would involve writing a backend for your frontend which would produce the appropriate D code, which could be a bit tricky, but not so bad if you're not worried about it producing the nicest-looking D code. Although the nice thing about this approach is that your DBasic frontend could be written in whatever you want, instead of being limited to C++ as with the other approach. In any case, it would be nice to have some kind of D grammar to start from, so you'd know what all the various language constructs your new syntax would have to support, rather than picking out all the grammar bits from the D spec. There _is_ one somewhere, but (1) I don't know if it's kept entirely up to date and (2) don't know what it's called or where it is. :|
Oct 30 2006
next sibling parent reply "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
"Jarrett Billingsley" <kb3ctd2 yahoo.com> wrote in message 
news:ei6a3b$2ffm$1 digitaldaemon.com...

 If this message that I'm replying to is rich text, then it looks through 
 my reader and on the web interface.
I meant to say that it looks *fine*. If a message were looking at me through my newsreader.. :S
Oct 30 2006
parent RA <ra wolven.net> writes:
Addendum to my previous post\ranting...

I can understand that back when computers were quite limited on memory,
processing
power, etc., that we had to make things as "efficient" as possible, thereby
semi-forcing humans to adapt to the machine.  Abbreviate everything, use
hieroglyphics and compression to save space, batch processing, utilize as little
as possible to get the job done...

But those days have been gone for a while now, and I think it's time we make the
machine adapt to the human way of looking at things.

Another thing that drive me nuts is Zero based indexing. That's just sick, wrong
and lazy on the lanuguage\compiler programmers part.  There is no such thing as
bucket\dog\tree number 0.  0 means NONE, notta, zip, zilch.  It DEFINATELY
doesn't
(and shouldn't) mean "the first" of a number of items...

Ahhh, I feel better now...  :)
Oct 31 2006
prev sibling parent reply RA <ra wolven.net> writes:
== Quote from Jarrett Billingsley
 What kind of syntax are you thinking?  (I'm genuinely interested :) )
 In any case, it would be nice to have some kind of D grammar to start from,
 so you'd know what all the various language constructs your new syntax would
 have to support, rather than picking out all the grammar bits from the D
 spec.  There _is_ one somewhere, but (1) I don't know if it's kept entirely
 up to date and (2) don't know what it's called or where it is.  :|
== Quote from Jarrett Billingsley The kind of syntax I'm thinking of would be one that would be easily readable and understood by virtually any programmer. It would be more verbose than C\C++\D\etc.. and cleaner and more logical than VB. Simple, readable, logical and consistent would be the primary goals. I would try to use generally understood words and symbols except where they just don't make sense unless you've been "educated" as to their meanings within a particular language. Obviously, that's not a completely achievable goal, but a worthwhile target in my opinion nonetheless. Here are a few examples. I think ==, &&, ||, etc. suck. = means equal, not ==. I would change the & and | symbols to something new. I would change the * to the "normal" multiplacation symbol (little x floating midline), the / to the normal divide symbol, the ! to the normal not equal (equal with a diagonal slash thru it, and I would use the ascii decimal 171 and 187 ((hex 00AB and 00BB) they look like small << and >>) as the "literal" identifiers instead of quotation marks... and so forth and so on. Things like "If a !<= b" just don't make any sense to me. I think it's much clearer and correct as "If a > b". Anyone that would use !<= is just being perverse IMO... :) I would use; Do, DoWhile, DoUntil, EndDo, EndDoWhile, EndDoUntil, If, EndIf, BegFunction, EndFunction, BegProp, EndProp, etc... I liked and applauded your "no need for '->'" comment in your VB example. I've been an application programmer for over 20 years. For the first 10 years I worked on the IBM S\36 and AS\400 using RPG. The next 10 years was programming Windows using Visual RPG (kind of a cross between freeform RPG and VB). One thing I've noticed over the years is that with each new generation of programming languages and operating systems, I keep getting moved closer and closer to system level programming. As a supposedly "high level" programmer, personally, I find that annoying. The fact that I "can" write my own string object is entertaining (if I felt like it), but the fact that I HAVE to absolutely sucks. Nice features that were built into the OS and language\compiler I used 15 years ago have dissappeared and now I have to do them MANUALLY. That sux. Memory management? Garbage collection? Ten years ago, I had no idea what that meant. I didn't NEED to because the OS and compiler took care of it for me... AUTOMATICALLY! God, what a novel concept. A few things I've noticed mentioned on the forums here... integer, double, float, unsigned integer, decimal, and on and on. WTH is that about? All I want is a NUMBER with nine digits and two decimals... you can store it in memory ANY DAMN WAY you like, but I don't NEED to know or care about it. And I damn sure don't want to have to manually convert "cast" it to make my math work. I think Walter mentioned that no one has a nice data file handling system. Other than the AS\400 and ASNA's duplication of that file system on Windows, I've never seen anything "nice". MSSQL, MySQL, etc... are a royal pain in the hiney, anything but nice. I've come to the conclusion that the only reason those things are used is because MOST programmers have never worked on a "nice" system with a "nice" OS, language and database. Anyway... enough ranting. :) I really like what I've read on here about Walters tossing of "legacy" compatibility and ignorant "features" of C++ when designing D. And trying to make it more modern with garbage collection, eliminating pointers (although it sure seems to have a lot of support for them), etc.. I just personally don't care for the syntax. But given my background, I think that's understandable. I think it would be great to give D a second syntax that would be "easier" to learn\understand for people that aren't coming from a C What do you think?
Oct 31 2006
parent "Jarrett Billingsley" <kb3ctd2 yahoo.com> writes:
"RA" <ra wolven.net> wrote in message 
news:ei8ij4$21ql$1 digitaldaemon.com...
 The kind of syntax I'm thinking of would be one that would be easily 
 readable and
 understood by virtually any programmer. It would be more verbose than
 C\C++\D\etc..  and cleaner and more  logical than VB.  Simple, readable, 
 logical
 and consistent would be the primary goals.  I would try to use generally
 understood words and symbols except where they just don't make sense 
 unless you've
 been "educated" as to their meanings within a particular language. 
 Obviously,
 that's not a completely achievable goal, but a worthwhile target in my 
 opinion
 nonetheless.
An admirable goal.
 Here are a few examples.  I think ==, &&, ||, etc. suck.  = means equal, 
 not ==.
 I would change the & and | symbols to something new.  I would change the * 
 to the
 "normal" multiplacation symbol (little x floating midline), the / to the 
 normal
 divide symbol, the ! to the normal not equal (equal with a diagonal slash 
 thru it,
 and I would use the ascii decimal 171 and 187 ((hex 00AB and 00BB) they 
 look like
 small << and >>) as the "literal" identifiers instead of quotation 
 marks... and so
 forth and so on.
I don't know how I'd type any of those, though. Now you're getting into APL territory ;)
 Things like "If a !<= b" just don't make any sense to me.  I think it's 
 much
 clearer and correct as "If a > b".  Anyone that would use !<= is just 
 being
 perverse IMO... :)
Actually !<= is not necessarily "not less than or equal" -- it's a special operator for comparing floating point numbers which takes into account NaNs. But again, typing the "not less than or equal" character on a keyboard is not very easy, nor is it probably supported under many fonts or under ASCII.
 I would use; Do, DoWhile, DoUntil, EndDo, EndDoWhile, EndDoUntil, If, 
 EndIf,
 BegFunction, EndFunction, BegProp, EndProp, etc...

 I liked and applauded your "no need for '->'" comment in your VB example.
Oh, I didn't make that -- that was in the original D example (have a look on the intro page of the spec). There's no -> in D either.
 I've been an application programmer for over 20 years.  For the first 10 
 years I
 worked on the IBM S\36 and AS\400 using RPG.  The next 10 years was 
 programming
 Windows using Visual RPG (kind of a cross between freeform RPG and VB).

 One thing I've noticed over the years is that with each new generation of
 programming languages and operating systems, I keep getting moved closer 
 and
 closer to system level programming.  As a supposedly "high level" 
 programmer,
 personally, I find that annoying.  The fact that I "can" write my own 
 string
 object is entertaining (if I felt like it), but the fact that I HAVE to 
 absolutely
 sucks.

 Nice features that were built into the OS and language\compiler I used 15 
 years
 ago have dissappeared and now I have to do them MANUALLY.  That sux. 
 Memory
 management? Garbage collection?  Ten years ago, I had no idea what that 
 meant.  I
 didn't NEED to because the OS and compiler took care of it for me...
 AUTOMATICALLY!  God, what a novel concept.
I think what's really happened is that there are still high-level languages, but low-level languages have come into their own so much that they're now popular for things like app development. Back in the late 70s, early 80s, what languages were there? You had the _really_ low level ones used for writing OSes, and then the much higher-level ones (like Smalltalk etc.) used for writing more interesting programs. But now the low-level languages have taken on many features from the high-level ones and have become more useful for things other than raw hardware access. There are still plenty of high-level languages that don't have explicit memory management or the need to program your own String object, e.g. Java,
 A few things I've noticed mentioned on the forums here... integer, double, 
 float,
 unsigned integer, decimal, and on and on.  WTH is that about?  All I want 
 is a
 NUMBER with nine digits and two decimals...  you can store it in memory 
 ANY DAMN
 WAY you like, but I don't NEED to know or care about it.  And I damn sure 
 don't
 want to have to manually convert "cast" it to make my math work.
Well D is a _systems_ programming language and not necessarily a numerical or financial programming language. As such, having precise control over the number of bits and format of any given number is crucial for proper interfacing with various APIs. An eleven-digit fixed-point number doesn't mean much to most OSes and libraries, and so having such a type in D would not make sense either.
 I think Walter mentioned that no one has a nice data file handling system. 
 Other
 than the AS\400 and ASNA's duplication of that file system on Windows, 
 I've never
 seen anything "nice".  MSSQL, MySQL, etc... are a royal pain in the hiney,
 anything but nice.  I've come to the conclusion that the only reason those 
 things
 are used is because MOST programmers have never worked on a "nice" system 
 with a
 "nice" OS, language and database.

 Anyway...  enough ranting.  :)

 I really like what I've read on here about Walters tossing of "legacy"
 compatibility and ignorant "features" of C++ when designing D.  And trying 
 to make
 it more modern with garbage collection, eliminating pointers (although it 
 sure
 seems to have a lot of support for them), etc..
Again, the support for pointers is because lots of systems code needs them. D thankfully provides enough other abstractions, however, so that most of the time you need not deal with them. About the only time I use pointers is with the associative array "in" operator and allocating structs on the heap, things which don't happen _too_ often.
 I just personally don't care for the syntax. But given my background, I 
 think
 that's understandable.  I think it would be great to give D a second 
 syntax that
 would be "easier" to learn\understand for people that aren't coming from a 
 C


 What do you think?
I think you're entirely justified in your opinion, and a language like that which you describe would probably be very useful to a lot of people. However, I don't know if D is really the right language to try to fit to your vision. You could change the syntax, but if you can't change the semantics, you won't be any closer to your dream. Have a go at making your own language then. It's fun, and that's from experience ;)
Nov 01 2006
prev sibling next sibling parent Hasan Aljudy <hasan.aljudy gmail.com> writes:
If you just want to create D with different syntax, I think you can 
write a stand-alone parser then use it convert your other-syntax code to 
D syntax.

RA wrote:
 I was curious about how hard it would be to create a different “language 
 front end” for D? 
 
 Since you (Walter) have made the compiler front end (lexer, parser, etc) 
 source code available,
 
 would it be possible for someone to modify it and create a different 
 (for example, something
 
 more like VB) language syntax that would still work with the back end 
 compiler?
 
  
 
 I realize most of you long time C\C++ programmers probably couldn’t see 
 any need for
 
 such a thing, but others of us (at least me) would prefer something not 
 quite as cryptic
 
 as C like syntax.
 
  
 
 I’ve never written a compiler\lexer\parser\etc. so I don’t know what all 
 would be involved
 
 in creating a different syntax.  In theory, it doesn’t seem like it 
 should be “all” that hard…
 
 But then, naiveté is great isn’t it.  J
 
Oct 30 2006
prev sibling parent BLS <nanali wanadoo.fr> writes:
To pick up Hasan's idea :
You can create a Translator from a BASIC like language to D without 
programming by using a tool named Depot4. (works quit nice to translate 
higher level languages 4GL / domain specific languages to lower level 
languages 3GL f.i. D)
All you have to do is do describe to source and target language in 
annotated EBNF.
A simple sample :


The Depot4 homepage :
http://www.math.tu-dresden.de/wir/depot4/

Good luck.
Björn
Oct 31 2006