www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.announce - CWrap - higher abstraction level for calling C functions

reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
D has complete (IMHO) compiler support for calling C functions (using 
extern(C)). But there is a lack of library support. Microsoft .NET 
Framework has such support, but it's poor (see previous thread about 
CWrap in digitalmars.D NG).

Once original function is properly described in IDL, CWrap gives the 
fallowing advantages:
* [memory-corruption-safe] User works with arrays, not separate pointers 
and lengths.
* [memory-leak-safe] Original function allocated memory will be freed on 
any failure (NOTE: see "Original function restrictions" in documentation).
* [fail-safe] An exception will be thrown if original function fails.
* [usability] Much easier to work with arrays/preferred string types (it 
converts UTF-8/16/32 back and forth, see docs).
* [speed] No unnecessary GC (or other) allocations/copying/reading.
* [garbage-free] Look at, e.g. std.file - it leaves lots of garbage 
because of calling native platform functions. Generated wrappers don't 
even allocate GC memory (except for output parameters when user do want 
get GC allocated buffers).

Sources: https://bitbucket.org/denis_sh/cwrap
Docs (incomplete): http://deoma-cmd.ru/d/docs/src/cwrap.html

Phobos is needed.
Besides Phobos, the only file needed for generating: cwrap.d
The only file needed for generated wrappers to work: internal/cwraprt.d

Library state: almost every internal function is unittested, generated 
wrappers pass rather complicated test-suit in internal/cwrapunittests.d

"Examples" is the most informative section of the documentation. Other 
examples:
1. Original functions, all in one file:
https://bitbucket.org/denis_sh/cwrap/src/tip/examples/c.d
https://bitbucket.org/denis_sh/cwrap/src/tip/internal/cwrapunittests.d
2. Generated wrappers:
https://bitbucket.org/denis_sh/cwrap/src/tip/out

NOTE1: You need dmd 2.058 to generate wrappers at CT.
NOTE2: You can change `enum data` to `auto data` in examples/c.d, this 
will reduce compilation time.

Previous thread about CWrap in digitalmars.D NG:
http://www.digitalmars.com/d/archives/digitalmars/D/Higher_abstraction_level_for_calling_C_functions_156633.html
Feb 20 2012
parent reply Gour <gour atmarama.net> writes:
On Mon, 20 Feb 2012 18:02:49 +0400
Denis Shelomovskij <verylonglogin.reg gmail.com> wrote:

Hello Denis,


 D has complete (IMHO) compiler support for calling C functions (using=20
 extern(C)). But there is a lack of library support.=20


I'm glad you're working on (another) bindings tool being aware that it's
not reasonable that D will cover all our library needs.

Otoh, I'm also concerned that using just wrappers over C code would
erode (potential) type-safety and advantage of using D in the first
place. (In our app we envision to use wXD for GUI, need bindings for 3rd
party C lib, using one of Sqlite3 wrappers...)


 Once original function is properly described in IDL, CWrap gives the=20
 fallowing advantages:


If we eliminate htod which is Windows only and considering that manual
wrapping is not very attractive option, along with your list which sound
quite goo, my question is what would be some of pro/cons of CWrap in
comparison with e.g. SWIG and it's D support as well as with dstep
(https://github.com/jacob-carlborg/dstep) if our main objective would be
to keep some of D's advantages such as having higher-level (D-ish) API,
exception handling (insted of error-checking), type-safety etc.
available in our D bindings of C lib(s)?


Sincerely,
Gour

--=20
One who restrains his senses, keeping them under full control,=20
and fixes his consciousness upon Me, is known as a man of=20
steady intelligence.

http://atmarama.net | Hlapicina (Croatia) | GPG: 52B5C810
Feb 20 2012
next sibling parent reply Jacob Carlborg <doob me.com> writes:
On 2012-02-20 16:34, Gour wrote:

 On Mon, 20 Feb 2012 18:02:49 +0400
 Denis Shelomovskij<verylonglogin.reg gmail.com>  wrote:

 Hello Denis,


 D has complete (IMHO) compiler support for calling C functions (using
 extern(C)). But there is a lack of library support.


 I'm glad you're working on (another) bindings tool being aware that it's
 not reasonable that D will cover all our library needs.

 Otoh, I'm also concerned that using just wrappers over C code would
 erode (potential) type-safety and advantage of using D in the first
 place. (In our app we envision to use wXD for GUI, need bindings for 3rd
 party C lib, using one of Sqlite3 wrappers...)


 Once original function is properly described in IDL, CWrap gives the
 fallowing advantages:


 If we eliminate htod which is Windows only and considering that manual
 wrapping is not very attractive option, along with your list which sound
 quite goo, my question is what would be some of pro/cons of CWrap in
 comparison with e.g. SWIG and it's D support as well as with dstep
 (https://github.com/jacob-carlborg/dstep) if our main objective would be
 to keep some of D's advantages such as having higher-level (D-ish) API,
 exception handling (insted of error-checking), type-safety etc.
 available in our D bindings of C lib(s)?


 Sincerely,
 Gour


As I understand it, CWrap provides more than just bindings for a C 
function. SWIG or dstep could output CWrap bindings.

-- 
/Jacob Carlborg
Feb 20 2012
parent reply Gour <gour atmarama.net> writes:
On Mon, 20 Feb 2012 20:52:04 +0100
Jacob Carlborg <doob me.com> wrote:


 As I understand it, CWrap provides more than just bindings for a C=20
 function.=20


Right, but, afaik, SWIG can also make more than wrapper for a C function
by using typemaps, %exception etc.

Does it mean that dstep is only tool for wrapping C function for usage
in D?


 SWIG or dstep could output CWrap bindings.


The former probably won't bother to do it, but wonder about the latter?


Sincerely,
Gour


--=20
Whatever action a great man performs, common men follow. And=20
whatever standards he sets by exemplary acts, all the world pursues.

http://atmarama.net | Hlapicina (Croatia) | GPG: 52B5C810
Feb 21 2012
parent reply Jacob Carlborg <doob me.com> writes:
On 2012-02-21 13:12, Gour wrote:

 On Mon, 20 Feb 2012 20:52:04 +0100
 Jacob Carlborg<doob me.com>  wrote:


 As I understand it, CWrap provides more than just bindings for a C
 function.


 Right, but, afaik, SWIG can also make more than wrapper for a C function
 by using typemaps, %exception etc.

 Does it mean that dstep is only tool for wrapping C function for usage
 in D?


 SWIG or dstep could output CWrap bindings.


 The former probably won't bother to do it, but wonder about the latter?


DStep is currently used only for automatically generating bindings for C 
functions and Objective-C classes and methods.

But it depends what code I output, I could change dstep to output code 
that uses CWrap.

-- 
/Jacob Carlborg
Feb 21 2012
next sibling parent reply Gour <gour atmarama.net> writes:
On Tue, 21 Feb 2012 18:40:20 +0100
Jacob Carlborg <doob me.com> wrote:


 DStep is currently used only for automatically generating bindings
 for C functions and Objective-C classes and methods.


OK. Thank you.


 But it depends what code I output, I could change dstep to output
 code that uses CWrap.


That would be nice and make the whole job of binding C functions and
providing customized D API easier...and let's not forget about extra
points for coordinating work. ;)


Sincerely,
Gour


--=20
One who is not connected with the Supreme can have neither=20
transcendental intelligence nor a steady mind, without which=20
there is no possibility of peace. And how can there be any=20
happiness without peace?

http://atmarama.net | Hlapicina (Croatia) | GPG: 52B5C810
Feb 22 2012
parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
22.02.2012 13:41, Gour пишет:

 That would be nice and make the whole job of binding C functions and
 providing customized D API easier...and let's not forget about extra
 points for coordinating work. ;)


I'm open for emails about using/changing/fixing CWrap.
Feb 22 2012
parent Jacob Carlborg <doob me.com> writes:
On 2012-02-22 19:59, Denis Shelomovskij wrote:

 22.02.2012 13:41, Gour пишет:

 That would be nice and make the whole job of binding C functions and
 providing customized D API easier...and let's not forget about extra
 points for coordinating work. ;)


 I'm open for emails about using/changing/fixing CWrap.


Before adding CWrap to DStep I need to finish the regular bindings 
generator.

-- 
/Jacob Carlborg
Feb 22 2012
prev sibling parent reply Gour <gour atmarama.net> writes:
On Tue, 21 Feb 2012 18:40:20 +0100
Jacob Carlborg <doob me.com> wrote:


 DStep is currently used only for automatically generating bindings
 for C functions and Objective-C classes and methods.


Do oyu find that dstep is capable to be used for binding whole C libs
instead of using SWIG?


Sincerely,
Gour


--=20
Everyone is forced to act helplessly according to the qualities=20
he has acquired from the modes of material nature; therefore no=20
one can refrain from doing something, not even for a moment.

http://atmarama.net | Hlapicina (Croatia) | GPG: 52B5C810
Mar 04 2012
parent reply Jacob Carlborg <doob me.com> writes:
On 2012-03-04 16:13, Gour wrote:

 On Tue, 21 Feb 2012 18:40:20 +0100
 Jacob Carlborg<doob me.com>  wrote:


 DStep is currently used only for automatically generating bindings
 for C functions and Objective-C classes and methods.


 Do oyu find that dstep is capable to be used for binding whole C libs
 instead of using SWIG?


 Sincerely,
 Gour


It's not there yet but that's the idea.

-- 
/Jacob Carlborg
Mar 04 2012
parent reply Gour <gour atmarama.net> writes:
On Sun, 04 Mar 2012 17:18:19 +0100
Jacob Carlborg <doob me.com> wrote:


 It's not there yet but that's the idea.


Thank you. It makes dstep attractive alternative then.

What would be some pro/cons of dstep vs SWIG?


Sincerely,
Gour


--=20
But a person free from all attachment and aversion and able=20
to control his senses through regulative principles of=20
freedom can obtain the complete mercy of the Lord.

http://atmarama.net | Hlapicina (Croatia) | GPG: 52B5C810
Mar 05 2012
parent reply Jacob Carlborg <doob me.com> writes:
On 2012-03-05 09:42, Gour wrote:

 On Sun, 04 Mar 2012 17:18:19 +0100
 Jacob Carlborg<doob me.com>  wrote:


 It's not there yet but that's the idea.


 Thank you. It makes dstep attractive alternative then.

 What would be some pro/cons of dstep vs SWIG?


 Sincerely,
 Gour


Well, I'm not familiar with SWIG but a few pros for DStep would probably be:

* Specifically targeted for D
* Handles Objective-C
* No need for interface files
* Uses a complete frontend to do the parsing
* Static linking

Pros for SWIG

* Handles C++
* In a more complete sate

-- 
/Jacob Carlborg
Mar 05 2012
next sibling parent Andrea Fontana <advmail katamail.com> writes:
Il giorno lun, 05/03/2012 alle 10.25 +0100, Jacob Carlborg ha scritto:


 Pros for SWIG
=20
 * Handles C++



And classes, of course!



 * In a more complete sate
Mar 05 2012
prev sibling parent reply Gour <gour atmarama.net> writes:
On Mon, 05 Mar 2012 10:25:28 +0100
Jacob Carlborg <doob me.com> wrote:


 * Specifically targeted for D


That's a good one.


 * Handles Objective-C


No interest for it atm.


 * No need for interface files


This is also nice, but I wonder if one can tailor D side of the binding
or dstep is meant to just provide wrapper for C function?

In the latter case, glueing dstsep with CWrape would be nice.


 * Uses a complete front-end to do the parsing
 * Static linking


Those are very nice as well...I might try dstep, so you can expect some
issues  github. ;)


 Pros for SWIG
=20
 * Handles C++


We anticipate that our C++ usage in D is limited just to GUI libs and
hopefully there will be new wxD soon.


 * In a more complete sate


That's true atm, but I'm not sure how much is klickverbot committed to
work on it...so, having specific binding tool for D is very nice in any
case.


Sincerely,
Gour

--=20
Mar 05 2012
parent Jacob Carlborg <doob me.com> writes:
On 2012-03-05 10:49, Gour wrote:

 On Mon, 05 Mar 2012 10:25:28 +0100
 Jacob Carlborg<doob me.com>  wrote:


 * Specifically targeted for D


 That's a good one.


 * Handles Objective-C


 No interest for it atm.


 * No need for interface files


 This is also nice, but I wonder if one can tailor D side of the binding
 or dstep is meant to just provide wrapper for C function?


Currently it's only meant to just provide bindings.


 In the latter case, glueing dstsep with CWrape would be nice.


 * Uses a complete front-end to do the parsing
 * Static linking


 Those are very nice as well...I might try dstep, so you can expect some
 issues  github. ;)


Unfortunately it's not in a state where it's ready to be tested yet. 
Sure you can always test it but there's no point in reporting issues for 
something I know isn't finished or working. But help developing the tool 
is always welcome.


 Pros for SWIG

 * Handles C++


 We anticipate that our C++ usage in D is limited just to GUI libs and
 hopefully there will be new wxD soon.


 * In a more complete sate


 That's true atm, but I'm not sure how much is klickverbot committed to
 work on it...so, having specific binding tool for D is very nice in any
 case.


 Sincerely,
 Gour



-- 
/Jacob Carlborg
Mar 05 2012
prev sibling parent reply Denis Shelomovskij <verylonglogin.reg gmail.com> writes:
20.02.2012 19:34, Gour пишет:

 On Mon, 20 Feb 2012 18:02:49 +0400
 Denis Shelomovskij<verylonglogin.reg gmail.com>  wrote:

 Hello Denis,


 D has complete (IMHO) compiler support for calling C functions (using
 extern(C)). But there is a lack of library support.


 I'm glad you're working on (another) bindings tool being aware that it's
 not reasonable that D will cover all our library needs.

 Otoh, I'm also concerned that using just wrappers over C code would
 erode (potential) type-safety and advantage of using D in the first
 place. (In our app we envision to use wXD for GUI, need bindings for 3rd
 party C lib, using one of Sqlite3 wrappers...)


 Once original function is properly described in IDL, CWrap gives the
 fallowing advantages:


 If we eliminate htod which is Windows only and considering that manual
 wrapping is not very attractive option, along with your list which sound
 quite goo, my question is what would be some of pro/cons of CWrap in
 comparison with e.g. SWIG and it's D support as well as with dstep
 (https://github.com/jacob-carlborg/dstep) if our main objective would be
 to keep some of D's advantages such as having higher-level (D-ish) API,
 exception handling (insted of error-checking), type-safety etc.
 available in our D bindings of C lib(s)?


 Sincerely,
 Gour


OK, I'm here again. Jacob Carlborg is right - CWrap is a different tool 
so it's incorrect to compare it with SWIG/dstep. It's developed as a 
high performance backend for other tools or for manual wrapping a couple 
of functions, not for translating entire *.h files or creating class 
hierarchy.

Let's imagine there is no std.file, no std.windows.charset, and no 
std.windows.registry. Look at my WinAPI examples - it will be easier and 
more robust to write IDL descriptions by hands (because you should read 
specs for every WinAPI function at least once) and use generated 
wrappers than to write lots of boring code to converts strings/check 
errors. E.g. I did it already (CWrap WinAPI examples) and it wasn't too 
long compared with reading specs.

Or let's imagine one will have to finally revise Phobos to stop 
producing garbage and/or make D usable without GC. He will kill himself 
if he want to write optimal-performance garbage-less calls to WinAPI or 
will spend lot of time by writing non-optimal. Or he will just use CWrap 
generated code.
Feb 22 2012
parent Gour <gour atmarama.net> writes:
On Wed, 22 Feb 2012 22:54:19 +0400
Denis Shelomovskij <verylonglogin.reg gmail.com> wrote:


 OK, I'm here again. Jacob Carlborg is right - CWrap is a different
 tool so it's incorrect to compare it with SWIG/dstep. It's developed
 as a high performance backend for other tools or for manual wrapping
 a couple of functions, not for translating entire *.h files or
 creating class hierarchy.


Thank you for clarification. The doc page just says: "Stuff for
automated wrapping functions written in C." So I thought it's meant for
binding whole libs.

However, I still believe that having dstep producing CWrap output would
be beneficial for those having task to provide D bindings for C libs
with customized D-ish API.


Sincerely,
Gour

--=20
But a person free from all attachment and aversion and able=20
to control his senses through regulative principles of=20
freedom can obtain the complete mercy of the Lord.

http://atmarama.net | Hlapicina (Croatia) | GPG: 52B5C810
Feb 22 2012