Re: [Lightning] [PATCH v2] Fix 9 tests on Windows x64, due to size of long

[Paulo César Pereira de Andrade]

Em qui, 3 de out de 2019 às 17:00, Zachary Cook
<address@hidden> escreveu:
>
> Here is the test log: 
> https://gist.github.com/ZachCook/c44142ee7c4b88ec319f0abb769c1e95

  Do you mean the problems happens with the suggested patch?

> This is running using a custom implementation of sys/mman, so that very well 
> could be the main issue, any hints as to what is broken would be helpful

  If it pass any test, there should not be any mmap related issues.

> On Thu, Oct 3, 2019 at 3:45 PM Zachary Cook <address@hidden> wrote:
>>
>> ---
>>  check/ccall.c     | 4 ++++
>>  check/lightning.c | 4 ++++
>>  2 files changed, 8 insertions(+)
>>
>> diff --git a/check/ccall.c b/check/ccall.c
>> index e454a32..7e15674 100644
>> --- a/check/ccall.c
>> +++ b/check/ccall.c
>> @@ -142,8 +142,12 @@ typedef unsigned short             _us;
>>  typedef signed int             _i;
>>  #if __WORDSIZE == 64
>>  typedef unsigned int           _ui;
>> +#if (__WIN32 && !__CYGWIN__)
>> +typedef signed long long       _l;
>> +#else
>>  typedef signed long            _l;
>>  #endif
>> +#endif
>>  typedef float                  _f;
>>  typedef double                 _d;
>>
>> diff --git a/check/lightning.c b/check/lightning.c
>> index 8df033a..b9cf431 100644
>> --- a/check/lightning.c
>> +++ b/check/lightning.c
>> @@ -74,6 +74,10 @@ static void                  *DL_HANDLE;
>>  #define PARSING_CODE           2
>>  #define MAX_IDENTIFIER         256
>>
>> +#if __WORDSIZE == 64 && (__WIN32 && !__CYGWIN__)
>> +#define long                   long long
>> +#endif
>> +
>>  /*
>>   * Types
>>   */
>> --
>> 2.23.0

  I believe I understand what you need. The proper solution should be
to basically replace all occurrences of 'long' with jit_word_t in
check/lightning.c,
and similar approach to check/ccall.c.

  I will do it, and git push shortly. After that, please let me know
if there are
any remaining issues.

  Are you using the plain MinGW compiler or some other environment?
I should be able to test with MinGW as well.

Thanks,
Paulo