# Non POSIX (really Windows) Support

**URL:** <https://discourse.ats-lang.org/t/non-posix-really-windows-support/676>\
**Category:** Google Group\
**Created:** [January 16, 2014, 8:14am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676 "2014-01-16T08:14:21Z")\
**Posts on this page:** 18\
**Page:** 1

<div class="post-metadata">

**Author:** ![H\_Zhang](https://avatars.discourse-cdn.com/v4/letter/h/b77776/32.png) [@H\_Zhang](https://discourse.ats-lang.org/u/H_Zhang)\
**Post date:** [January 16, 2014, 8:14am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/1 "2014-01-16T08:14:21Z")

</div>

Hi,

I have been trying to compile ATS2 using mingw (under Cygwin it is no  
problem) and I have run into some roadblocks mostly related to POSIX  
assumptions. I was able to work around the superfluous ones by commenting  
out most POSIX types related things (like uid/gid etc. that does not appear  
to be used by the compiler itself anyway). However at the final link stage  
there are errors related to sigsetjmp and siglongjmp:

pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference to  
`siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to`siglongjmp’  
pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to`sigsetjmp’  
/usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
collect2: error: ld returned 1 exit status  
make[1]: \*\*\* [patsopt] Error 1  
make: \*\*\* [src2\_patsopt] Error 2

These appear to be used by the exception mechanism.

Is there any plan to make ATS2 compatible with Windows? Limiting to Cygwin  
could make it hard for non-open source usages.

Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [March 24, 2014, 4:59pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/2 "2014-03-24T16:59:28Z")

</div>

Then it should be largely straightforward. Essentially, you need  
Windows-equivalent of the following package:

ATS2-Postiats-include-0.0.6.tgz[http://sourceforge.net/projects/ats2-lang/files/ats2-lang/ats2-postiats-0.0.6/ATS2-Postiats-include-0.0.6.tgz/download](http://sourceforge.net/projects/ats2-lang/files/ats2-lang/ats2-postiats-0.0.6/ATS2-Postiats-include-0.0.6.tgz/download)

which is available in the following directory:

> **[ats2-lang - Browse /ats2-lang/ats2-postiats-0.0.6 at SourceForge.net](https://sourceforge.net/projects/ats2-lang/files/ats2-lang/ats2-postiats-0.0.6/)**
>
> ATS2: Unleashing the potentials of types and templates

I can set up a project to create such a package for Windows.On Monday, March 24, 2014 4:33:51 AM UTC-4, Cyrille Duret wrote:

> I just mean about the generated code could run under window.  
> Ideal would be same as Rust language to have proper support os mingw/msys  
> solution under windows that would generate native windows executables
> 
> [GitHub - rust-lang/rust: Empowering everyone to build reliable and efficient software.](https://github.com/mozilla/rust/wiki/Note-getting-started-developing-Rust#windows)
> 
> Surely I would think about pushing ATS in my big projects if I had better  
> support of all major platforms
> 
> cheers  
> 😉
> 
> Le dimanche 23 mars 2014 16:28:35 UTC+1, gmhwxi a écrit :
> 
> > Do you mean (1), or (2), or both?
> > 
> > 1. To make sure that the ATS compiler (ATS/Positats) can run under windows
> > 2. To make sure that the generated C code from ATS source can run under  
> > windows
> > 
> > On Sunday, March 23, 2014 4:41:49 AM UTC-4, Cyrille Duret wrote:
> > 
> > > Would it be so hard work to make a really independant implementation of  
> > > ATS and POSIX so one could use it for native windows apps ?
> > > 
> > > I think ATS is a wonderful language but because of that no one will use  
> > > it for a real project because of lack of a proper windows support.
> > > 
> > > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > > 
> > > > That is good to know. Cheers!
> > > > 
> > > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > > 
> > > > > 1. The tip helped. It allowed me to move forward.
> > > > > 
> > > > > 2. Still I had to get around more errors related mostly to the file  
> > > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > > version does not seem to function correctly and I suspect the problem is  
> > > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > > 
> > > > > 3. For now I think the more promising path is to keep the tool chain  
> > > > > in Cygwin and cross compile to mingw, that way one can focus on addressing  
> > > > > the incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5  
> > > > > release and compiled under Cygwin. I then tested cross compiling to mingw.  
> > > > > It seems to work as long as one avoid incompatible headers (such as  
> > > > > types.cats, unistd, fcntl, wait, stat, mman and maybe more 🙂 are not  
> > > > > pulled in. Cross compiled exe can run standalone in Windows. Cool!
> > > > > 
> > > > > Haitao
> > > > > 
> > > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > > 
> > > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > > 
> > > > > > Is there support for longjmp and setjmp under Windows?
> > > > > > 
> > > > > > If there is, please do:
> > > > > > 
> > > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > > 
> > > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > > 
> > > > > > > Hi,
> > > > > > > 
> > > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is  
> > > > > > > no problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > > 
> > > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined  
> > > > > > > reference to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > > collect2: error: ld returned 1 exit status  
> > > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > > 
> > > > > > > These appear to be used by the exception mechanism.
> > > > > > > 
> > > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > > 
> > > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![H\_Zhang](https://avatars.discourse-cdn.com/v4/letter/h/b77776/32.png) [@H\_Zhang](https://discourse.ats-lang.org/u/H_Zhang)\
**Post date:** [March 24, 2014, 12:51am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/3 "2014-03-24T00:51:39Z")

</div>

Since interfacing with C in ATS is as simple as modeling the function  
prototypes and C structs in ATS it is doubtful that this is what is  
holding up ATS.On Sunday, March 23, 2014 1:41:49 AM UTC-7, Cyrille Duret wrote:

> Would it be so hard work to make a really independant implementation of  
> ATS and POSIX so one could use it for native windows apps ?
> 
> I think ATS is a wonderful language but because of that no one will use it  
> for a real project because of lack of a proper windows support.
> 
> Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> 
> > That is good to know. Cheers!
> > 
> > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > 
> > > 1. The tip helped. It allowed me to move forward.
> > > 
> > > 2. Still I had to get around more errors related mostly to the file  
> > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > version does not seem to function correctly and I suspect the problem is  
> > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > 
> > > 3. For now I think the more promising path is to keep the tool chain in  
> > > Cygwin and cross compile to mingw, that way one can focus on addressing the  
> > > incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5 release  
> > > and compiled under Cygwin. I then tested cross compiling to mingw. It seems  
> > > to work as long as one avoid incompatible headers (such as types.cats,  
> > > unistd, fcntl, wait, stat, mman and maybe more 🙂 are not pulled in. Cross  
> > > compiled exe can run standalone in Windows. Cool!
> > > 
> > > Haitao
> > > 
> > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > 
> > > > There is no need to use siglongjmp and sigsetjmp.
> > > > 
> > > > Is there support for longjmp and setjmp under Windows?
> > > > 
> > > > If there is, please do:
> > > > 
> > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > 
> > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > 
> > > > > Hi,
> > > > > 
> > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > > > > problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > 
> > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference  
> > > > > to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > collect2: error: ld returned 1 exit status  
> > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > 
> > > > > These appear to be used by the exception mechanism.
> > > > > 
> > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > Cygwin could make it hard for non-open source usages.
> > > > > 
> > > > > Haitao

---

<div class="post-metadata">

**Author:** ![Cyrille\_Duret](https://avatars.discourse-cdn.com/v4/letter/c/3d9bf3/32.png) [@Cyrille\_Duret](https://discourse.ats-lang.org/u/Cyrille_Duret)\
**Post date:** [March 24, 2014, 6:08pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/4 "2014-03-24T18:08:08Z")

</div>

Cool !  
As I see ATS2 is already able to generate clean native windows executable.  
The hard work is indeed to make the postiats-include package for windows.  
It would be nice if you could create an empty package for windows so we  
could start to begin some experiments.

I understand better the workflow now :

- generate the C files under cygwin,
- crosscompile it to mingw with the win32-postiats-include package

My concern is at this time networking support for all platform and I look  
at libuv to make a minimal embedded http server  
I think I will make a minimal server (based on this project  
[GitHub - philips/libuv-webserver: simple webserver in libuv](https://github.com/philips/libuv-webserver)) and try to compile it on all  
platforms.

cheers ;-)Le lundi 24 mars 2014 18:08:46 UTC+1, gmhwxi a écrit :

> > > I would agree step 3) could use some help to have wrapper files make  
> > > using Windows as convenient as POSIX but it is just as well to be driven by  
> > > a project that actually uses ATS.
> 
> Yes, I would say that project-driven development for this sort of thing is  
> the best.
> 
> On Monday, March 24, 2014 12:55:35 PM UTC-4, H Zhang wrote:
> 
> > Cyrille,
> > 
> > As I have documented earlier in this thread it is not hard to get Windows  
> > executables:
> > 
> > 1. You need to install Cygwin and the MinGW cross compiler from the  
> > Cygwin installer;
> > 2. You can build ATS itself under Cygwin;
> > 3. Make sure your ATS user program does not load library files that are  
> > POSIX dependent; if you need Windows functions, model the prototype in ATS  
> > following the book chapter on working with C code;
> > 4. Run cross compiler on the generated C code to generate Windows  
> > executables
> > 
> > ATS is not a huge system and it is fairly straightforward to figure out  
> > what to do if you run into problems. And if you can’t figure it out on your  
> > own people on this list will be more than happy to help. If you have a  
> > project to work through the kinks please publish you experience so others  
> > can learn too. Early adopters are expected to contribute towards the  
> > success of ATS!
> > 
> > I would agree step 3) could use some help to have wrapper files make  
> > using Windows as convenient as POSIX but it is just as well to be driven by  
> > a project that actually uses ATS.
> > 
> > Haitao
> > 
> > On Monday, March 24, 2014 1:33:51 AM UTC-7, Cyrille Duret wrote:
> > 
> > > I just mean about the generated code could run under window.  
> > > Ideal would be same as Rust language to have proper support os  
> > > mingw/msys solution under windows that would generate native windows  
> > > executables
> > > 
> > > [GitHub - rust-lang/rust: Empowering everyone to build reliable and efficient software.](https://github.com/mozilla/rust/wiki/Note-getting-started-developing-Rust#windows)
> > > 
> > > Surely I would think about pushing ATS in my big projects if I had  
> > > better support of all major platforms
> > > 
> > > cheers  
> > > 😉
> > > 
> > > Le dimanche 23 mars 2014 16:28:35 UTC+1, gmhwxi a écrit :
> > > 
> > > > Do you mean (1), or (2), or both?
> > > > 
> > > > 1. To make sure that the ATS compiler (ATS/Positats) can run under  
> > > > windows
> > > > 2. To make sure that the generated C code from ATS source can run under  
> > > > windows
> > > > 
> > > > On Sunday, March 23, 2014 4:41:49 AM UTC-4, Cyrille Duret wrote:
> > > > 
> > > > > Would it be so hard work to make a really independant implementation  
> > > > > of ATS and POSIX so one could use it for native windows apps ?
> > > > > 
> > > > > I think ATS is a wonderful language but because of that no one will  
> > > > > use it for a real project because of lack of a proper windows support.
> > > > > 
> > > > > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > > > > 
> > > > > > That is good to know. Cheers!
> > > > > > 
> > > > > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > > > > 
> > > > > > > 1. The tip helped. It allowed me to move forward.
> > > > > > > 
> > > > > > > 2. Still I had to get around more errors related mostly to the file  
> > > > > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > > > > version does not seem to function correctly and I suspect the problem is  
> > > > > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > > > > 
> > > > > > > 3. For now I think the more promising path is to keep the tool chain  
> > > > > > > in Cygwin and cross compile to mingw, that way one can focus on addressing  
> > > > > > > the incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5  
> > > > > > > release and compiled under Cygwin. I then tested cross compiling to mingw.  
> > > > > > > It seems to work as long as one avoid incompatible headers (such as  
> > > > > > > types.cats, unistd, fcntl, wait, stat, mman and maybe more 🙂 are not  
> > > > > > > pulled in. Cross compiled exe can run standalone in Windows. Cool!
> > > > > > > 
> > > > > > > Haitao
> > > > > > > 
> > > > > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > > > > 
> > > > > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > > > > 
> > > > > > > > Is there support for longjmp and setjmp under Windows?
> > > > > > > > 
> > > > > > > > If there is, please do:
> > > > > > > > 
> > > > > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > > > > 
> > > > > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > > > > 
> > > > > > > > > Hi,
> > > > > > > > > 
> > > > > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is  
> > > > > > > > > no problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > > > > 
> > > > > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined  
> > > > > > > > > reference to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > > > > collect2: error: ld returned 1 exit status  
> > > > > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > > > > 
> > > > > > > > > These appear to be used by the exception mechanism.
> > > > > > > > > 
> > > > > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting  
> > > > > > > > > to Cygwin could make it hard for non-open source usages.
> > > > > > > > > 
> > > > > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [January 16, 2014, 9:53pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/5 "2014-01-16T21:53:46Z")

</div>

That is good to know. Cheers!On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:

> 1. The tip helped. It allowed me to move forward.
> 
> 2. Still I had to get around more errors related mostly to the file  
> systems and processes to get patsopt to compile. Unfortunately the compiled  
> version does not seem to function correctly and I suspect the problem is  
> with POSIX file handling (stat vs lstat, some path issues etc).
> 
> 3. For now I think the more promising path is to keep the tool chain in  
> Cygwin and cross compile to mingw, that way one can focus on addressing the  
> incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5 release  
> and compiled under Cygwin. I then tested cross compiling to mingw. It seems  
> to work as long as one avoid incompatible headers (such as types.cats,  
> unistd, fcntl, wait, stat, mman and maybe more 🙂 are not pulled in. Cross  
> compiled exe can run standalone in Windows. Cool!
> 
> Haitao
> 
> On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> 
> > There is no need to use siglongjmp and sigsetjmp.
> > 
> > Is there support for longjmp and setjmp under Windows?
> > 
> > If there is, please do:
> > 
> > #define sigsetjmp(env, savesigs) setjmp(env)  
> > #define siglongjmp(env, val) longjmp(env, val)
> > 
> > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > 
> > > Hi,
> > > 
> > > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > > problem) and I have run into some roadblocks mostly related to POSIX  
> > > assumptions. I was able to work around the superfluous ones by commenting  
> > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > to be used by the compiler itself anyway). However at the final link stage  
> > > there are errors related to sigsetjmp and siglongjmp:
> > > 
> > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference to  
> > > `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > collect2: error: ld returned 1 exit status  
> > > make[1]: \*\*\* [patsopt] Error 1  
> > > make: \*\*\* [src2\_patsopt] Error 2
> > > 
> > > These appear to be used by the exception mechanism.
> > > 
> > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > Cygwin could make it hard for non-open source usages.
> > > 
> > > Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [March 24, 2014, 5:32am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/6 "2014-03-24T05:32:14Z")

</div>

> > The early adopters will likely be more looking for performance (memory,  
> > latency and speed) and productivity  
> > (needs to be better than C to be attractive). I also think  
> > implementing library in ATS is more likely than  
> > sizable standalone applications completely written in ATS, at least in  
> > the beginning.

I completely agree. I have high hope for the template system of ATS2, which  
is primarily designed to greatly increase  
programming productivity.

> > even if you can prove that your code should work there is still no  
> > guarantee that ATS will translate it correctly

Compiler errors are fundamentally different from errors introduced by  
programmers.  
ATS compiler is a ‘closed world’; it can get better after being used more  
and more. While a programmer can get  
better, the programmer cannot write code forever.On Monday, March 24, 2014 1:10:12 AM UTC-4, H Zhang wrote:

> Dependent types/linear types do seem formidable to use. It is not even  
> easy to understand code base as mature as the ATS library. On the other  
> hand ATS does seem to generate quality C code and can effortlessly  
> interface with C and work with unboxed/C data with no overhead. That is  
> what is attractive to me.
> 
> It is hard to imagine that early adopters of ATS will be those who are  
> mainly concerned with correctness due to the very limited usage history of  
> ATS – even if you can prove that your code should work there is still no  
> guarantee that ATS will translate it correctly (I am not questioning ATS’s  
> implementation quality just that seems to be the way how people think). The  
> early adopters will likely be more looking for performance (memory, latency  
> and speed) and productivity (needs to be better than C to be attractive). I  
> also think implementing library in ATS is more likely than sizable  
> standalone applications completely written in ATS, at least in the  
> beginning. As I noted before GC dependence is a problem for implementing  
> library in ATS though there are mitigation techniques that we have  
> discussed, including using multiple heaps (which would suit my requirement  
> for now).
> 
> Haitao
> 
> On Sunday, March 23, 2014 8:13:04 PM UTC-7, gmhwxi wrote:
> 
> > I have to say that ATS is difficult to use in general. My impression is  
> > that  
> > everyone (including myself) trying to use ATS has struggled a lot. The  
> > learning  
> > curve for ATS, as of now, may be simply too steep for most of programmers.
> > 
> > However, I do believe that there are programmers out there who will,  
> > somehow and someday,  
> > be able to make effective use of ATS in writing high-quality software.
> > 
> > On Sunday, March 23, 2014 8:51:39 PM UTC-4, H Zhang wrote:
> > 
> > > Since interfacing with C in ATS is as simple as modeling the function  
> > > prototypes and C structs in ATS it is doubtful that this is what is  
> > > holding up ATS.
> > > 
> > > On Sunday, March 23, 2014 1:41:49 AM UTC-7, Cyrille Duret wrote:
> > > 
> > > > Would it be so hard work to make a really independant implementation of  
> > > > ATS and POSIX so one could use it for native windows apps ?
> > > > 
> > > > I think ATS is a wonderful language but because of that no one will use  
> > > > it for a real project because of lack of a proper windows support.
> > > > 
> > > > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > > > 
> > > > > That is good to know. Cheers!
> > > > > 
> > > > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > > > 
> > > > > > 1. The tip helped. It allowed me to move forward.
> > > > > > 
> > > > > > 2. Still I had to get around more errors related mostly to the file  
> > > > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > > > version does not seem to function correctly and I suspect the problem is  
> > > > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > > > 
> > > > > > 3. For now I think the more promising path is to keep the tool chain  
> > > > > > in Cygwin and cross compile to mingw, that way one can focus on addressing  
> > > > > > the incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5  
> > > > > > release and compiled under Cygwin. I then tested cross compiling to mingw.  
> > > > > > It seems to work as long as one avoid incompatible headers (such as  
> > > > > > types.cats, unistd, fcntl, wait, stat, mman and maybe more 🙂 are not  
> > > > > > pulled in. Cross compiled exe can run standalone in Windows. Cool!
> > > > > > 
> > > > > > Haitao
> > > > > > 
> > > > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > > > 
> > > > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > > > 
> > > > > > > Is there support for longjmp and setjmp under Windows?
> > > > > > > 
> > > > > > > If there is, please do:
> > > > > > > 
> > > > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > > > 
> > > > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > > > 
> > > > > > > > Hi,
> > > > > > > > 
> > > > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is  
> > > > > > > > no problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > > > 
> > > > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined  
> > > > > > > > reference to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > > > collect2: error: ld returned 1 exit status  
> > > > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > > > 
> > > > > > > > These appear to be used by the exception mechanism.
> > > > > > > > 
> > > > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > > > 
> > > > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [March 24, 2014, 3:13am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/7 "2014-03-24T03:13:04Z")

</div>

I have to say that ATS is difficult to use in general. My impression is that  
everyone (including myself) trying to use ATS has struggled a lot. The  
learning  
curve for ATS, as of now, may be simply too steep for most of programmers.

However, I do believe that there are programmers out there who will,  
somehow and someday,  
be able to make effective use of ATS in writing high-quality software.On Sunday, March 23, 2014 8:51:39 PM UTC-4, H Zhang wrote:

> Since interfacing with C in ATS is as simple as modeling the function  
> prototypes and C structs in ATS it is doubtful that this is what is  
> holding up ATS.
> 
> On Sunday, March 23, 2014 1:41:49 AM UTC-7, Cyrille Duret wrote:
> 
> > Would it be so hard work to make a really independant implementation of  
> > ATS and POSIX so one could use it for native windows apps ?
> > 
> > I think ATS is a wonderful language but because of that no one will use  
> > it for a real project because of lack of a proper windows support.
> > 
> > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > 
> > > That is good to know. Cheers!
> > > 
> > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > 
> > > > 1. The tip helped. It allowed me to move forward.
> > > > 
> > > > 2. Still I had to get around more errors related mostly to the file  
> > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > version does not seem to function correctly and I suspect the problem is  
> > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > 
> > > > 3. For now I think the more promising path is to keep the tool chain in  
> > > > Cygwin and cross compile to mingw, that way one can focus on addressing the  
> > > > incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5 release  
> > > > and compiled under Cygwin. I then tested cross compiling to mingw. It seems  
> > > > to work as long as one avoid incompatible headers (such as types.cats,  
> > > > unistd, fcntl, wait, stat, mman and maybe more 🙂 are not pulled in. Cross  
> > > > compiled exe can run standalone in Windows. Cool!
> > > > 
> > > > Haitao
> > > > 
> > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > 
> > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > 
> > > > > Is there support for longjmp and setjmp under Windows?
> > > > > 
> > > > > If there is, please do:
> > > > > 
> > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > 
> > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > 
> > > > > > Hi,
> > > > > > 
> > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > > > > > problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > 
> > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference  
> > > > > > to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > collect2: error: ld returned 1 exit status  
> > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > 
> > > > > > These appear to be used by the exception mechanism.
> > > > > > 
> > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > 
> > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [January 16, 2014, 8:45am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/8 "2014-01-16T08:45:16Z")

</div>

There is no need to use siglongjmp and sigsetjmp.

Is there support for longjmp and setjmp under Windows?

If there is, please do:

#define sigsetjmp(env, savesigs) setjmp(env)  
#define siglongjmp(env, val) longjmp(env, val)On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:

> Hi,
> 
> I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> problem) and I have run into some roadblocks mostly related to POSIX  
> assumptions. I was able to work around the superfluous ones by commenting  
> out most POSIX types related things (like uid/gid etc. that does not appear  
> to be used by the compiler itself anyway). However at the final link stage  
> there are errors related to sigsetjmp and siglongjmp:
> 
> pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference to  
> `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> collect2: error: ld returned 1 exit status  
> make[1]: \*\*\* [patsopt] Error 1  
> make: \*\*\* [src2\_patsopt] Error 2
> 
> These appear to be used by the exception mechanism.
> 
> Is there any plan to make ATS2 compatible with Windows? Limiting to Cygwin  
> could make it hard for non-open source usages.
> 
> Haitao

---

<div class="post-metadata">

**Author:** ![Zhiqiang\_Ren](https://avatars.discourse-cdn.com/v4/letter/z/9fc29f/32.png) [@Zhiqiang\_Ren](https://discourse.ats-lang.org/u/Zhiqiang_Ren)\
**Post date:** [March 24, 2014, 4:27pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/9 "2014-03-24T16:27:47Z")

</div>

I don’t think it’s that important to make ATS compiler to run on all major  
platforms like Windows. Maybe it’s more interesting to make an ATS compiler  
running on Linux to generate C code which is compilabe and runnable on  
Windows.On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:

> Hi,
> 
> I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> problem) and I have run into some roadblocks mostly related to POSIX  
> assumptions. I was able to work around the superfluous ones by commenting  
> out most POSIX types related things (like uid/gid etc. that does not appear  
> to be used by the compiler itself anyway). However at the final link stage  
> there are errors related to sigsetjmp and siglongjmp:
> 
> pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference to  
> `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> collect2: error: ld returned 1 exit status  
> make[1]: \*\*\* [patsopt] Error 1  
> make: \*\*\* [src2\_patsopt] Error 2
> 
> These appear to be used by the exception mechanism.
> 
> Is there any plan to make ATS2 compatible with Windows? Limiting to Cygwin  
> could make it hard for non-open source usages.
> 
> Haitao

---

<div class="post-metadata">

**Author:** ![H\_Zhang](https://avatars.discourse-cdn.com/v4/letter/h/b77776/32.png) [@H\_Zhang](https://discourse.ats-lang.org/u/H_Zhang)\
**Post date:** [March 24, 2014, 4:55pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/10 "2014-03-24T16:55:35Z")

</div>

Cyrille,

As I have documented earlier in this thread it is not hard to get Windows  
executables:

1. You need to install Cygwin and the MinGW cross compiler from the Cygwin  
installer;
2. You can build ATS itself under Cygwin;
3. Make sure your ATS user program does not load library files that are  
POSIX dependent; if you need Windows functions, model the prototype in ATS  
following the book chapter on working with C code;
4. Run cross compiler on the generated C code to generate Windows  
executables

ATS is not a huge system and it is fairly straightforward to figure out  
what to do if you run into problems. And if you can’t figure it out on your  
own people on this list will be more than happy to help. If you have a  
project to work through the kinks please publish you experience so others  
can learn too. Early adopters are expected to contribute towards the  
success of ATS!

I would agree step 3) could use some help to have wrapper files make using  
Windows as convenient as POSIX but it is just as well to be driven by a  
project that actually uses ATS.

HaitaoOn Monday, March 24, 2014 1:33:51 AM UTC-7, Cyrille Duret wrote:

> I just mean about the generated code could run under window.  
> Ideal would be same as Rust language to have proper support os mingw/msys  
> solution under windows that would generate native windows executables
> 
> [GitHub - rust-lang/rust: Empowering everyone to build reliable and efficient software.](https://github.com/mozilla/rust/wiki/Note-getting-started-developing-Rust#windows)
> 
> Surely I would think about pushing ATS in my big projects if I had better  
> support of all major platforms
> 
> cheers  
> 😉
> 
> Le dimanche 23 mars 2014 16:28:35 UTC+1, gmhwxi a écrit :
> 
> > Do you mean (1), or (2), or both?
> > 
> > 1. To make sure that the ATS compiler (ATS/Positats) can run under windows
> > 2. To make sure that the generated C code from ATS source can run under  
> > windows
> > 
> > On Sunday, March 23, 2014 4:41:49 AM UTC-4, Cyrille Duret wrote:
> > 
> > > Would it be so hard work to make a really independant implementation of  
> > > ATS and POSIX so one could use it for native windows apps ?
> > > 
> > > I think ATS is a wonderful language but because of that no one will use  
> > > it for a real project because of lack of a proper windows support.
> > > 
> > > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > > 
> > > > That is good to know. Cheers!
> > > > 
> > > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > > 
> > > > > 1. The tip helped. It allowed me to move forward.
> > > > > 
> > > > > 2. Still I had to get around more errors related mostly to the file  
> > > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > > version does not seem to function correctly and I suspect the problem is  
> > > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > > 
> > > > > 3. For now I think the more promising path is to keep the tool chain  
> > > > > in Cygwin and cross compile to mingw, that way one can focus on addressing  
> > > > > the incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5  
> > > > > release and compiled under Cygwin. I then tested cross compiling to mingw.  
> > > > > It seems to work as long as one avoid incompatible headers (such as  
> > > > > types.cats, unistd, fcntl, wait, stat, mman and maybe more 🙂 are not  
> > > > > pulled in. Cross compiled exe can run standalone in Windows. Cool!
> > > > > 
> > > > > Haitao
> > > > > 
> > > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > > 
> > > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > > 
> > > > > > Is there support for longjmp and setjmp under Windows?
> > > > > > 
> > > > > > If there is, please do:
> > > > > > 
> > > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > > 
> > > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > > 
> > > > > > > Hi,
> > > > > > > 
> > > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is  
> > > > > > > no problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > > 
> > > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined  
> > > > > > > reference to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > > collect2: error: ld returned 1 exit status  
> > > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > > 
> > > > > > > These appear to be used by the exception mechanism.
> > > > > > > 
> > > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > > 
> > > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [March 24, 2014, 5:08pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/11 "2014-03-24T17:08:46Z")

</div>

> > I would agree step 3) could use some help to have wrapper files make  
> > using Windows as convenient as POSIX but it is just as well to be driven by  
> > a project that actually uses ATS.

Yes, I would say that project-driven development for this sort of thing is  
the best.On Monday, March 24, 2014 12:55:35 PM UTC-4, H Zhang wrote:

> Cyrille,
> 
> As I have documented earlier in this thread it is not hard to get Windows  
> executables:
> 
> 1. You need to install Cygwin and the MinGW cross compiler from the Cygwin  
> installer;
> 2. You can build ATS itself under Cygwin;
> 3. Make sure your ATS user program does not load library files that are  
> POSIX dependent; if you need Windows functions, model the prototype in ATS  
> following the book chapter on working with C code;
> 4. Run cross compiler on the generated C code to generate Windows  
> executables
> 
> ATS is not a huge system and it is fairly straightforward to figure out  
> what to do if you run into problems. And if you can’t figure it out on your  
> own people on this list will be more than happy to help. If you have a  
> project to work through the kinks please publish you experience so others  
> can learn too. Early adopters are expected to contribute towards the  
> success of ATS!
> 
> I would agree step 3) could use some help to have wrapper files make using  
> Windows as convenient as POSIX but it is just as well to be driven by a  
> project that actually uses ATS.
> 
> Haitao
> 
> On Monday, March 24, 2014 1:33:51 AM UTC-7, Cyrille Duret wrote:
> 
> > I just mean about the generated code could run under window.  
> > Ideal would be same as Rust language to have proper support os mingw/msys  
> > solution under windows that would generate native windows executables
> > 
> > [GitHub - rust-lang/rust: Empowering everyone to build reliable and efficient software.](https://github.com/mozilla/rust/wiki/Note-getting-started-developing-Rust#windows)
> > 
> > Surely I would think about pushing ATS in my big projects if I had better  
> > support of all major platforms
> > 
> > cheers  
> > 😉
> > 
> > Le dimanche 23 mars 2014 16:28:35 UTC+1, gmhwxi a écrit :
> > 
> > > Do you mean (1), or (2), or both?
> > > 
> > > 1. To make sure that the ATS compiler (ATS/Positats) can run under  
> > > windows
> > > 2. To make sure that the generated C code from ATS source can run under  
> > > windows
> > > 
> > > On Sunday, March 23, 2014 4:41:49 AM UTC-4, Cyrille Duret wrote:
> > > 
> > > > Would it be so hard work to make a really independant implementation of  
> > > > ATS and POSIX so one could use it for native windows apps ?
> > > > 
> > > > I think ATS is a wonderful language but because of that no one will use  
> > > > it for a real project because of lack of a proper windows support.
> > > > 
> > > > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > > > 
> > > > > That is good to know. Cheers!
> > > > > 
> > > > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > > > 
> > > > > > 1. The tip helped. It allowed me to move forward.
> > > > > > 
> > > > > > 2. Still I had to get around more errors related mostly to the file  
> > > > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > > > version does not seem to function correctly and I suspect the problem is  
> > > > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > > > 
> > > > > > 3. For now I think the more promising path is to keep the tool chain  
> > > > > > in Cygwin and cross compile to mingw, that way one can focus on addressing  
> > > > > > the incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5  
> > > > > > release and compiled under Cygwin. I then tested cross compiling to mingw.  
> > > > > > It seems to work as long as one avoid incompatible headers (such as  
> > > > > > types.cats, unistd, fcntl, wait, stat, mman and maybe more 🙂 are not  
> > > > > > pulled in. Cross compiled exe can run standalone in Windows. Cool!
> > > > > > 
> > > > > > Haitao
> > > > > > 
> > > > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > > > 
> > > > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > > > 
> > > > > > > Is there support for longjmp and setjmp under Windows?
> > > > > > > 
> > > > > > > If there is, please do:
> > > > > > > 
> > > > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > > > 
> > > > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > > > 
> > > > > > > > Hi,
> > > > > > > > 
> > > > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is  
> > > > > > > > no problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > > > 
> > > > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined  
> > > > > > > > reference to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > > > collect2: error: ld returned 1 exit status  
> > > > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > > > 
> > > > > > > > These appear to be used by the exception mechanism.
> > > > > > > > 
> > > > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > > > 
> > > > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [March 23, 2014, 3:28pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/12 "2014-03-23T15:28:35Z")

</div>

Do you mean (1), or (2), or both?

1. To make sure that the ATS compiler (ATS/Positats) can run under windows
2. To make sure that the generated C code from ATS source can run under  
windowsOn Sunday, March 23, 2014 4:41:49 AM UTC-4, Cyrille Duret wrote:

> Would it be so hard work to make a really independant implementation of  
> ATS and POSIX so one could use it for native windows apps ?
> 
> I think ATS is a wonderful language but because of that no one will use it  
> for a real project because of lack of a proper windows support.
> 
> Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> 
> > That is good to know. Cheers!
> > 
> > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > 
> > > 1. The tip helped. It allowed me to move forward.
> > > 
> > > 2. Still I had to get around more errors related mostly to the file  
> > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > version does not seem to function correctly and I suspect the problem is  
> > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > 
> > > 3. For now I think the more promising path is to keep the tool chain in  
> > > Cygwin and cross compile to mingw, that way one can focus on addressing the  
> > > incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5 release  
> > > and compiled under Cygwin. I then tested cross compiling to mingw. It seems  
> > > to work as long as one avoid incompatible headers (such as types.cats,  
> > > unistd, fcntl, wait, stat, mman and maybe more 🙂 are not pulled in. Cross  
> > > compiled exe can run standalone in Windows. Cool!
> > > 
> > > Haitao
> > > 
> > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > 
> > > > There is no need to use siglongjmp and sigsetjmp.
> > > > 
> > > > Is there support for longjmp and setjmp under Windows?
> > > > 
> > > > If there is, please do:
> > > > 
> > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > 
> > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > 
> > > > > Hi,
> > > > > 
> > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > > > > problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > 
> > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference  
> > > > > to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > collect2: error: ld returned 1 exit status  
> > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > 
> > > > > These appear to be used by the exception mechanism.
> > > > > 
> > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > Cygwin could make it hard for non-open source usages.
> > > > > 
> > > > > Haitao

---

<div class="post-metadata">

**Author:** ![Brandon\_Barker](https://sea1.discourse-cdn.com/flex019/user_avatar/discourse.ats-lang.org/brandon_barker/32/61_2.png) [@Brandon\_Barker](https://discourse.ats-lang.org/u/Brandon_Barker)\
**Post date:** [March 24, 2014, 4:47pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/13 "2014-03-24T16:47:57Z")

</div>

(or even running in cygwin, since that seems to work fine for now)

Brandon Barker  
brandon…@gmail.comOn Mon, Mar 24, 2014 at 12:27 PM, Zhiqiang Ren [alex.r...@gmail.com](mailto:alex.r...@gmail.com)wrote:

> I don’t think it’s that important to make ATS compiler to run on all major  
> platforms like Windows. Maybe it’s more interesting to make an ATS compiler  
> running on Linux to generate C code which is compilabe and runnable on  
> Windows.
> 
> On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> 
> > Hi,
> > 
> > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > problem) and I have run into some roadblocks mostly related to POSIX  
> > assumptions. I was able to work around the superfluous ones by commenting  
> > out most POSIX types related things (like uid/gid etc. that does not appear  
> > to be used by the compiler itself anyway). However at the final link stage  
> > there are errors related to sigsetjmp and siglongjmp:
> > 
> > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference to  
> > `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > collect2: error: ld returned 1 exit status  
> > make[1]: \*\*\* [patsopt] Error 1  
> > make: \*\*\* [src2\_patsopt] Error 2
> > 
> > These appear to be used by the exception mechanism.
> > 
> > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > Cygwin could make it hard for non-open source usages.
> > 
> > Haitao
> 
> –  
> You received this message because you are subscribed to the Google Groups  
> “ats-lang-users” group.  
> To unsubscribe from this group and stop receiving emails from it, send an  
> email to [ats-lang-user...@googlegroups.com](mailto:ats-lang-user...@googlegroups.com).  
> To post to this group, send email to [ats-lan...@googlegroups.com](mailto:ats-lan...@googlegroups.com).  
> To view this discussion on the web visit  
> [https://groups.google.com/d/msgid/ats-lang-users/0e8891e2-c9ef-4033-96aa-ed68abf2579c%40googlegroups.com](https://groups.google.com/d/msgid/ats-lang-users/0e8891e2-c9ef-4033-96aa-ed68abf2579c%40googlegroups.com)[https://groups.google.com/d/msgid/ats-lang-users/0e8891e2-c9ef-4033-96aa-ed68abf2579c%40googlegroups.com?utm\_medium=email&utm\_source=footer](https://groups.google.com/d/msgid/ats-lang-users/0e8891e2-c9ef-4033-96aa-ed68abf2579c%40googlegroups.com?utm_medium=email&utm_source=footer)  
> .

---

<div class="post-metadata">

**Author:** ![H\_Zhang](https://avatars.discourse-cdn.com/v4/letter/h/b77776/32.png) [@H\_Zhang](https://discourse.ats-lang.org/u/H_Zhang)\
**Post date:** [January 16, 2014, 9:03pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/14 "2014-01-16T21:03:01Z")

</div>

1. The tip helped. It allowed me to move forward.

2. Still I had to get around more errors related mostly to the file systems  
and processes to get patsopt to compile. Unfortunately the compiled version  
does not seem to function correctly and I suspect the problem is with POSIX  
file handling (stat vs lstat, some path issues etc).

3. For now I think the more promising path is to keep the tool chain in  
Cygwin and cross compile to mingw, that way one can focus on addressing the  
incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5 release  
and compiled under Cygwin. I then tested cross compiling to mingw. It seems  
to work as long as one avoid incompatible headers (such as types.cats,  
unistd, fcntl, wait, stat, mman and maybe more 🙂 are not pulled in. Cross  
compiled exe can run standalone in Windows. Cool!

HaitaoOn Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:

> There is no need to use siglongjmp and sigsetjmp.
> 
> Is there support for longjmp and setjmp under Windows?
> 
> If there is, please do:
> 
> #define sigsetjmp(env, savesigs) setjmp(env)  
> #define siglongjmp(env, val) longjmp(env, val)
> 
> On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> 
> > Hi,
> > 
> > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > problem) and I have run into some roadblocks mostly related to POSIX  
> > assumptions. I was able to work around the superfluous ones by commenting  
> > out most POSIX types related things (like uid/gid etc. that does not appear  
> > to be used by the compiler itself anyway). However at the final link stage  
> > there are errors related to sigsetjmp and siglongjmp:
> > 
> > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference to  
> > `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > collect2: error: ld returned 1 exit status  
> > make[1]: \*\*\* [patsopt] Error 1  
> > make: \*\*\* [src2\_patsopt] Error 2
> > 
> > These appear to be used by the exception mechanism.
> > 
> > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > Cygwin could make it hard for non-open source usages.
> > 
> > Haitao

---

<div class="post-metadata">

**Author:** ![H\_Zhang](https://avatars.discourse-cdn.com/v4/letter/h/b77776/32.png) [@H\_Zhang](https://discourse.ats-lang.org/u/H_Zhang)\
**Post date:** [March 24, 2014, 5:10am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/15 "2014-03-24T05:10:12Z")

</div>

Dependent types/linear types do seem formidable to use. It is not even easy  
to understand code base as mature as the ATS library. On the other hand ATS  
does seem to generate quality C code and can effortlessly interface with C  
and work with unboxed/C data with no overhead. That is what is attractive  
to me.

It is hard to imagine that early adopters of ATS will be those who are  
mainly concerned with correctness due to the very limited usage history of  
ATS – even if you can prove that your code should work there is still no  
guarantee that ATS will translate it correctly (I am not questioning ATS’s  
implementation quality just that seems to be the way how people think). The  
early adopters will likely be more looking for performance (memory, latency  
and speed) and productivity (needs to be better than C to be attractive). I  
also think implementing library in ATS is more likely than sizable  
standalone applications completely written in ATS, at least in the  
beginning. As I noted before GC dependence is a problem for implementing  
library in ATS though there are mitigation techniques that we have  
discussed, including using multiple heaps (which would suit my requirement  
for now).

HaitaoOn Sunday, March 23, 2014 8:13:04 PM UTC-7, gmhwxi wrote:

> I have to say that ATS is difficult to use in general. My impression is  
> that  
> everyone (including myself) trying to use ATS has struggled a lot. The  
> learning  
> curve for ATS, as of now, may be simply too steep for most of programmers.
> 
> However, I do believe that there are programmers out there who will,  
> somehow and someday,  
> be able to make effective use of ATS in writing high-quality software.
> 
> On Sunday, March 23, 2014 8:51:39 PM UTC-4, H Zhang wrote:
> 
> > Since interfacing with C in ATS is as simple as modeling the function  
> > prototypes and C structs in ATS it is doubtful that this is what is  
> > holding up ATS.
> > 
> > On Sunday, March 23, 2014 1:41:49 AM UTC-7, Cyrille Duret wrote:
> > 
> > > Would it be so hard work to make a really independant implementation of  
> > > ATS and POSIX so one could use it for native windows apps ?
> > > 
> > > I think ATS is a wonderful language but because of that no one will use  
> > > it for a real project because of lack of a proper windows support.
> > > 
> > > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > > 
> > > > That is good to know. Cheers!
> > > > 
> > > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > > 
> > > > > 1. The tip helped. It allowed me to move forward.
> > > > > 
> > > > > 2. Still I had to get around more errors related mostly to the file  
> > > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > > version does not seem to function correctly and I suspect the problem is  
> > > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > > 
> > > > > 3. For now I think the more promising path is to keep the tool chain  
> > > > > in Cygwin and cross compile to mingw, that way one can focus on addressing  
> > > > > the incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5  
> > > > > release and compiled under Cygwin. I then tested cross compiling to mingw.  
> > > > > It seems to work as long as one avoid incompatible headers (such as  
> > > > > types.cats, unistd, fcntl, wait, stat, mman and maybe more 🙂 are not  
> > > > > pulled in. Cross compiled exe can run standalone in Windows. Cool!
> > > > > 
> > > > > Haitao
> > > > > 
> > > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > > 
> > > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > > 
> > > > > > Is there support for longjmp and setjmp under Windows?
> > > > > > 
> > > > > > If there is, please do:
> > > > > > 
> > > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > > 
> > > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > > 
> > > > > > > Hi,
> > > > > > > 
> > > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is  
> > > > > > > no problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > > 
> > > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined  
> > > > > > > reference to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > > collect2: error: ld returned 1 exit status  
> > > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > > 
> > > > > > > These appear to be used by the exception mechanism.
> > > > > > > 
> > > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > > 
> > > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![gmhwxi](https://avatars.discourse-cdn.com/v4/letter/g/67e7ee/32.png) [@gmhwxi](https://discourse.ats-lang.org/u/gmhwxi)\
**Post date:** [March 24, 2014, 5:05pm UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/16 "2014-03-24T17:05:07Z")

</div>

By the way, for file-related operations, I suggest that SDL2 be  
used. SDL2 has a very liberal licence; it is for developing games  
that are OS-independent, but you can certainly use it for other  
purposes.

–Hongwei

PS: I haven’t looked at the support for threads in SDL2, but I have  
a feeling that it is also good for developing OS-independent software  
in general.On Monday, March 24, 2014 12:59:28 PM UTC-4, gmhwxi wrote:

> Then it should be largely straightforward. Essentially, you need  
> Windows-equivalent of the following package:
> 
> ATS2-Postiats-include-0.0.6.tgz[http://sourceforge.net/projects/ats2-lang/files/ats2-lang/ats2-postiats-0.0.6/ATS2-Postiats-include-0.0.6.tgz/download](http://sourceforge.net/projects/ats2-lang/files/ats2-lang/ats2-postiats-0.0.6/ATS2-Postiats-include-0.0.6.tgz/download)
> 
> which is available in the following directory:
> 
> [ats2-lang - Browse /ats2-lang/ats2-postiats-0.0.6 at SourceForge.net](http://sourceforge.net/projects/ats2-lang/files/ats2-lang/ats2-postiats-0.0.6/)
> 
> I can set up a project to create such a package for Windows.
> 
> On Monday, March 24, 2014 4:33:51 AM UTC-4, Cyrille Duret wrote:
> 
> > I just mean about the generated code could run under window.  
> > Ideal would be same as Rust language to have proper support os mingw/msys  
> > solution under windows that would generate native windows executables
> > 
> > [GitHub - rust-lang/rust: Empowering everyone to build reliable and efficient software.](https://github.com/mozilla/rust/wiki/Note-getting-started-developing-Rust#windows)
> > 
> > Surely I would think about pushing ATS in my big projects if I had better  
> > support of all major platforms
> > 
> > cheers  
> > 😉
> > 
> > Le dimanche 23 mars 2014 16:28:35 UTC+1, gmhwxi a écrit :
> > 
> > > Do you mean (1), or (2), or both?
> > > 
> > > 1. To make sure that the ATS compiler (ATS/Positats) can run under  
> > > windows
> > > 2. To make sure that the generated C code from ATS source can run under  
> > > windows
> > > 
> > > On Sunday, March 23, 2014 4:41:49 AM UTC-4, Cyrille Duret wrote:
> > > 
> > > > Would it be so hard work to make a really independant implementation of  
> > > > ATS and POSIX so one could use it for native windows apps ?
> > > > 
> > > > I think ATS is a wonderful language but because of that no one will use  
> > > > it for a real project because of lack of a proper windows support.
> > > > 
> > > > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > > > 
> > > > > That is good to know. Cheers!
> > > > > 
> > > > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > > > 
> > > > > > 1. The tip helped. It allowed me to move forward.
> > > > > > 
> > > > > > 2. Still I had to get around more errors related mostly to the file  
> > > > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > > > version does not seem to function correctly and I suspect the problem is  
> > > > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > > > 
> > > > > > 3. For now I think the more promising path is to keep the tool chain  
> > > > > > in Cygwin and cross compile to mingw, that way one can focus on addressing  
> > > > > > the incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5  
> > > > > > release and compiled under Cygwin. I then tested cross compiling to mingw.  
> > > > > > It seems to work as long as one avoid incompatible headers (such as  
> > > > > > types.cats, unistd, fcntl, wait, stat, mman and maybe more 🙂 are not  
> > > > > > pulled in. Cross compiled exe can run standalone in Windows. Cool!
> > > > > > 
> > > > > > Haitao
> > > > > > 
> > > > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > > > 
> > > > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > > > 
> > > > > > > Is there support for longjmp and setjmp under Windows?
> > > > > > > 
> > > > > > > If there is, please do:
> > > > > > > 
> > > > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > > > 
> > > > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > > > 
> > > > > > > > Hi,
> > > > > > > > 
> > > > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is  
> > > > > > > > no problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > > > 
> > > > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined  
> > > > > > > > reference to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > > > collect2: error: ld returned 1 exit status  
> > > > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > > > 
> > > > > > > > These appear to be used by the exception mechanism.
> > > > > > > > 
> > > > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > > > 
> > > > > > > > Haitao

---

<div class="post-metadata">

**Author:** ![Cyrille\_Duret](https://avatars.discourse-cdn.com/v4/letter/c/3d9bf3/32.png) [@Cyrille\_Duret](https://discourse.ats-lang.org/u/Cyrille_Duret)\
**Post date:** [March 23, 2014, 8:41am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/17 "2014-03-23T08:41:49Z")

</div>

Would it be so hard work to make a really independant implementation of ATS  
and POSIX so one could use it for native windows apps ?

I think ATS is a wonderful language but because of that no one will use it  
for a real project because of lack of a proper windows support.Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :

> That is good to know. Cheers!
> 
> On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> 
> > 1. The tip helped. It allowed me to move forward.
> > 
> > 2. Still I had to get around more errors related mostly to the file  
> > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > version does not seem to function correctly and I suspect the problem is  
> > with POSIX file handling (stat vs lstat, some path issues etc).
> > 
> > 3. For now I think the more promising path is to keep the tool chain in  
> > Cygwin and cross compile to mingw, that way one can focus on addressing the  
> > incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5 release  
> > and compiled under Cygwin. I then tested cross compiling to mingw. It seems  
> > to work as long as one avoid incompatible headers (such as types.cats,  
> > unistd, fcntl, wait, stat, mman and maybe more 🙂 are not pulled in. Cross  
> > compiled exe can run standalone in Windows. Cool!
> > 
> > Haitao
> > 
> > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > 
> > > There is no need to use siglongjmp and sigsetjmp.
> > > 
> > > Is there support for longjmp and setjmp under Windows?
> > > 
> > > If there is, please do:
> > > 
> > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > #define siglongjmp(env, val) longjmp(env, val)
> > > 
> > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > 
> > > > Hi,
> > > > 
> > > > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > > > problem) and I have run into some roadblocks mostly related to POSIX  
> > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > there are errors related to sigsetjmp and siglongjmp:
> > > > 
> > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference  
> > > > to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > collect2: error: ld returned 1 exit status  
> > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > 
> > > > These appear to be used by the exception mechanism.
> > > > 
> > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > Cygwin could make it hard for non-open source usages.
> > > > 
> > > > Haitao

---

<div class="post-metadata">

**Author:** ![Cyrille\_Duret](https://avatars.discourse-cdn.com/v4/letter/c/3d9bf3/32.png) [@Cyrille\_Duret](https://discourse.ats-lang.org/u/Cyrille_Duret)\
**Post date:** [March 24, 2014, 8:33am UTC](https://discourse.ats-lang.org/t/non-posix-really-windows-support/676/18 "2014-03-24T08:33:51Z")

</div>

I just mean about the generated code could run under window.  
Ideal would be same as Rust language to have proper support os mingw/msys  
solution under windows that would generate native windows executables

> **[GitHub - rust-lang/rust: Empowering everyone to build reliable and efficient...](https://github.com/rust-lang/rust#windows)**
>
> Empowering everyone to build reliable and efficient software. - GitHub - rust-lang/rust: Empowering everyone to build reliable and efficient software.

Surely I would think about pushing ATS in my big projects if I had better  
support of all major platforms

cheers  
;-)Le dimanche 23 mars 2014 16:28:35 UTC+1, gmhwxi a écrit :

> Do you mean (1), or (2), or both?
> 
> 1. To make sure that the ATS compiler (ATS/Positats) can run under windows
> 2. To make sure that the generated C code from ATS source can run under  
> windows
> 
> On Sunday, March 23, 2014 4:41:49 AM UTC-4, Cyrille Duret wrote:
> 
> > Would it be so hard work to make a really independant implementation of  
> > ATS and POSIX so one could use it for native windows apps ?
> > 
> > I think ATS is a wonderful language but because of that no one will use  
> > it for a real project because of lack of a proper windows support.
> > 
> > Le jeudi 16 janvier 2014 22:53:46 UTC+1, gmhwxi a écrit :
> > 
> > > That is good to know. Cheers!
> > > 
> > > On Thursday, January 16, 2014 4:03:01 PM UTC-5, H Zhang wrote:
> > > 
> > > > 1. The tip helped. It allowed me to move forward.
> > > > 
> > > > 2. Still I had to get around more errors related mostly to the file  
> > > > systems and processes to get patsopt to compile. Unfortunately the compiled  
> > > > version does not seem to function correctly and I suspect the problem is  
> > > > with POSIX file handling (stat vs lstat, some path issues etc).
> > > > 
> > > > 3. For now I think the more promising path is to keep the tool chain in  
> > > > Cygwin and cross compile to mingw, that way one can focus on addressing the  
> > > > incompatibilities in ATS2 libraries. I downloaded the newest 0.0.5 release  
> > > > and compiled under Cygwin. I then tested cross compiling to mingw. It seems  
> > > > to work as long as one avoid incompatible headers (such as types.cats,  
> > > > unistd, fcntl, wait, stat, mman and maybe more 🙂 are not pulled in. Cross  
> > > > compiled exe can run standalone in Windows. Cool!
> > > > 
> > > > Haitao
> > > > 
> > > > On Thursday, January 16, 2014 12:45:16 AM UTC-8, gmhwxi wrote:
> > > > 
> > > > > There is no need to use siglongjmp and sigsetjmp.
> > > > > 
> > > > > Is there support for longjmp and setjmp under Windows?
> > > > > 
> > > > > If there is, please do:
> > > > > 
> > > > > #define sigsetjmp(env, savesigs) setjmp(env)  
> > > > > #define siglongjmp(env, val) longjmp(env, val)
> > > > > 
> > > > > On Thursday, January 16, 2014 3:14:21 AM UTC-5, H Zhang wrote:
> > > > > 
> > > > > > Hi,
> > > > > > 
> > > > > > I have been trying to compile ATS2 using mingw (under Cygwin it is no  
> > > > > > problem) and I have run into some roadblocks mostly related to POSIX  
> > > > > > assumptions. I was able to work around the superfluous ones by commenting  
> > > > > > out most POSIX types related things (like uid/gid etc. that does not appear  
> > > > > > to be used by the compiler itself anyway). However at the final link stage  
> > > > > > there are errors related to sigsetjmp and siglongjmp:
> > > > > > 
> > > > > > pats\_error\_dats.o:pats\_error\_dats.c:(.text+0x22): undefined reference  
> > > > > > to `siglongjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0x1e): undefined reference to `siglongjmp’  
> > > > > > pats\_staexp2\_skexp\_dats.o:pats\_staexp2\_skexp\_dats.c:(.text+0xcf7):  
> > > > > > undefined reference to `sigsetjmp' pats_staexp2_skexp_dats.o:pats_staexp2_skexp_dats.c:(.text+0xd97): undefined reference to `sigsetjmp’  
> > > > > > /usr/lib/gcc/x86\_64-w64-mingw32/4.8.1/…/…/…/…/x86\_64-w64-mingw32/bin/ld:  
> > > > > > pats\_staexp2\_skexp\_dats.o: bad reloc address 0x0 in section `.pdata’  
> > > > > > collect2: error: ld returned 1 exit status  
> > > > > > make[1]: \*\*\* [patsopt] Error 1  
> > > > > > make: \*\*\* [src2\_patsopt] Error 2
> > > > > > 
> > > > > > These appear to be used by the exception mechanism.
> > > > > > 
> > > > > > Is there any plan to make ATS2 compatible with Windows? Limiting to  
> > > > > > Cygwin could make it hard for non-open source usages.
> > > > > > 
> > > > > > Haitao