remove the c++ syntax, struct::xxx

[jiangzidong]

jerry now has the c++ syntax, struct_name:xxx and enum_name:xxx
such as sizeof (ecma_collection_chunk_t::data) and mem_block_length_type_t::GENERAL
those are not standard c syntax.

I convert them to JERRY_SIZE_OF_STRUCT_MEMBER (ecma_collection_chunk_t, data) and GENERAL

JERRY_SIZE_OF_STRUCT_MEMBER is a macro defined in jrt.h

[rtakacs]

Please do not use a hack for getting the size. Instead, you should use macros.

 lit_utf8_byte_t data[ sizeof (uint64_t) - sizeof (mem_cpointer_t) ];

For example:

#define ECMA_COLLECTION_DATA_COUNT (sizeof (uint64_t) - sizeof (mem_cpointer_t))
#define ECMA_COLLECTION_DATA_SIZE (sizeof (lit_utf8_byte_t) * ECMA_COLLECTION_DATA_COUNT)

lit_utf8_byte_t data[ ECMA_COLLECTION_DATA_COUNT ];

const size_t values_in_chunk = ECMA_COLLECTION_DATA_SIZE / sizeof (ecma_value_t);

As I wrote (#806 (comment)) I'm already eliminating the C++ features. So the current workflow is not the best because it results work duplication. The example that I show is my result for this problem.

[rtakacs]

Could we talk on the jerryscript gitter about this topic? We can share our experience and current status :)

[jiangzidong]

@rtakacs Should I close this PR and let you continue C-Convertion work? Of course we can talk on gitter. What about your 13:00 (GMT+1), which is my 20:00(GMT+8) today?

[rtakacs]

@jiangzidong It is not necessary. You spent time to do this like me, but you were faster to upload. Your solution is also good (the null pointer surprised me, I have never seen similar before). In my comment I just would like to symbolize that either work is unnecessary when we do this together (at the same time) without any conversation.

The time you recommended is good for me.

[jiangzidong]

@rtakacs I don't think it's a hack. it is a well accepted method to get size of c struct member. (see http://stackoverflow.com/questions/3553296/c-sizeof-single-struct-member and http://stackoverflow.com/questions/3864583/sizeof-a-struct-member)

your solution directly calculated the size of ecma_collection_chunk_t::data

[akosthekiss]

@jiangzidong It's not that nice to repeat the sizeof-null-deref logic all over the code. It'd be better to have it in a macro, as suggested in the linked stackoverflow post, e.g.,

#define member_size(type, member) sizeof(((type *)0)->member)

(If we choose to use this approach. There are some comments that it is either an undefined behaviour in old C, or needs C99 compiler at least.)

[jiangzidong]

@akiss77 @rtakacs I find that there already has the marco in jrt.h
line 191 #define JERRY_SIZE_OF_STRUCT_MEMBER(struct_name, member_name) sizeof (((struct_name *)NULL)->member_name)

I will fix my patch soon

[ruben-ayrapetyan]

@jiangzidong, please rename GENERAL / ONE_CHUNKED to MEM_BLOCK_LENGTH_TYPE_GENERAL / MEM_BLOCK_LENGTH_TYPE_ONE_CHUNKED.

[jiangzidong]

@ruben-ayrapetyan updated, thanks

[ruben-ayrapetyan]

@jiangzidong, please, fix indentation in the comment.

[jiangzidong]

@ruben-ayrapetyan fixed

[ruben-ayrapetyan]

Looks good to me

[zherczeg]

LGTM

[LaszloLango]

Landed (9ab24b3)