This is the mail archive of the cygwin-developers mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: RFC: Cygwin 64 bit?


On 19 January 2012 10:25, Corinna Vinschen wrote:
> On Jan 18 15:54, Corinna Vinschen wrote:
>> On Jan 18 14:23, Pedro Alves wrote:
>> > I was looking for a technical reason that's worse, other then the
>> > amount of mechanical work, which I'm sure could also be easily
>> > scripted, with the long/int hidden behind a #define,
>> > from the perspective of code that uses the headers. Â It's obvious
>> > the pragma is less intrusive to the headers.
>>
>> I can't think of any. ÂIn both cases you will get the type mismatches in
>> certain circumstances within the LP64 code of the application.
>>
>> Hmm. ÂThe only other problem I could see with the #pragma approach is a
>> potential clash of debug information within the same CU. ÂIf you use
>> Windows headers you would potentially have two entries for long in
>> .debug_info:
>>
>> ÂFrom the LLP64 code:
>>
>> Â Â<1><d0>: Abbrev Number: 2 (DW_TAG_base_type)
>>    <d1>  DW_AT_byte_size  : 4
>>    <d2>  DW_AT_encoding  Â: 5    Â(signed)
>>    <d3>  DW_AT_name    Â: long int
>>
>> ÂFrom the LP64 code in the same CU:
>>
>> Â Â<1><dc>: Abbrev Number: 2 (DW_TAG_base_type)
>>    <dd>  DW_AT_byte_size  : 8
>>    <de>  DW_AT_encoding  Â: 5    Â(signed)
>>    <df>  DW_AT_name    Â: long int
>
> That won't be the case, fortunately. ÂA collegue has created a POC patch
> which would result in only one entry for long, the "right" one with
> DW_AT_byte_size 8, while the LLP64 usage of long would be converted to
> the type int internally.

Great. Does that also avoid nonsensical error messages along the lines
of "Cannot assign 'long *' to 'long *'"?

Andy


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]