Re: [Tinycc-devel] SEH in tcc

[PerfectDark]

Thank you for your request! I will necessarily check out tinycx.

Now I try to understand tcc code and, at first, try to divide it into small and 
independent components: tcc_file(file i/o), tcc_utils(tcc helper functions such 
as dynarray_add and so on), tcc_memory(all memory ops), tcc_scan(lexer+parser), 
tcc_backend(code generation). I added support to 'virtual' files, i.e. files, 
that internally added to tcc (it was need for me), so there is no need to have 
separate libtcc1.a. I debugged tcc under msvc with msvc' crtdbg helpers and I 
found that memory allocated in tcc.c::__sym_malloc is never released.
If I have enough time, I'll finish my work.
I think, that it would be great - to have _stable_, fast and ISO-conformant C 
compiler.

>On 5/20/06, PerfectDark <address@hidden> wrote:
>>
>> I work with tcc under win32 and now appeared one question - is there any
>> 'cheap' possibility to implement win32 MSVC-like SEH (__try, __except,
>> __finally, __leave) in tcc? Does someone interested in that, except me? May
>> be someone heard about any tcc' development plans - if them exists?
>> As I understood - in answers to my previous question, there are no any
>> branches of developing/testing tcc :(
>
>
> I've been working on some "extensions" to C using tcc as a base. The
>website is http://www.tinycx.org. I think there are other project that take
>tcc and experiment on it. In terms of SEH what I've done with tinycx is
>added some simple error handling with the reserved labels error: and
>finally:. The implementation uses setjmp/longjmp but that's because it was
>easier for me. Other features to C I've added are module support, multiple
>return values, function overloading, polymorphism, type inference. If you
>check it out any feedback you have would be appreciated.
>
>-Ben