Is there any update to alignment/padding for structs now that we can write 
.so files that can be called by C programs? We would have to know it to 
tell the C compiler what to expect from go side imho. 

On Thursday, April 5, 2012 at 9:10:48 PM UTC-7, Kyle Lemons wrote:
> On Thu, Apr 5, 2012 at 7:11 AM, LRN < <javascript:>> 
> wrote:
>> Related question:
>> If alignment and padding is platform dependent, how do you write and read 
>> structs to/from a socket?
> The solution is to not write platform dependent code.  The related topic 
> of byte ordering and platform independence is covered in this blog post 
> <>.  The gob 
> package <> (incidentally, if I'm not 
> mistaken, also written by the author of that blog post), is a performant 
> binary encoding scheme that will work from and to any platform that 
> supports go.
>> For example, in C it is not uncommon to declare a struct, initialize its 
>> fields with values (converted to big-endian format), then write it into a 
>> socket in one call.
>> The receiver will, again, read the struct from a socket in one call, then 
>> will do big-endian-to-native format conversion on each field where it is 
>> applicable - and will get the same struct the sender had.
>> This only works when definitions of the struct on both the sender and the 
>> receiver are aligned in the same way (usually - 1-byte aligned).
>> I've encountered this problem recently, which is why i'm asking.
>> Also, MSVC uses a padding algorithm different from the one gcc has, which 
>> is why there's a -mms-bitfields option in gcc. It forces gcc to use 
>> MSVC-style padding, which is required if the caller wants to pass certain 
>> structs to W32 API.
>> How will that work in Go? (h-m-m-m...i can probably infer that from W32 
>> port of Go by reading its source code...)

You received this message because you are subscribed to the Google Groups 
"golang-nuts" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
For more options, visit

Reply via email to