Make C++ compatible #13
Conversation
Signed-off-by: Jorge Perez <jjperez@ekumenlabs.com>
Signed-off-by: Jorge Perez <jjperez@ekumenlabs.com>
Signed-off-by: Jorge Perez <jjperez@ekumenlabs.com>
Signed-off-by: Jorge Perez <jjperez@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
|
Alright, @Snaipe PTAL! |
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
hidmic
force-pushed
the
hidmic/cxx-compatibility
branch
from
1858560 to
077c389
Compare
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
hidmic
force-pushed
the
hidmic/cxx-compatibility
branch
2 times, most recently
from
14c28eb to
a82593e
Compare
* Target external, undefined symbols only. * Fix indirect symbol type check. * Fix table lookup. Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
hidmic
force-pushed
the
hidmic/cxx-compatibility
branch
2 times, most recently
from
335b1be to
46afad4
Compare
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
hidmic
force-pushed
the
hidmic/cxx-compatibility
branch
2 times, most recently
from
2665884 to
da21a47
Compare
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
hidmic
force-pushed
the
hidmic/cxx-compatibility
branch
from
da21a47 to
fb0e5b3
Compare
Signed-off-by: Michel Hidalgo <michel@ekumenlabs.com>
| }; | ||
|
|
||
| typedef void(*mmk_fn)(void); | ||
| typedef struct mmk_stub *mmk_stub; | ||
| struct mmk_stub; |
There was a problem hiding this comment.
This breaks the API as far as I'm concerned. I'd rather have mmk_stub be an alias for struct mmk_stub *. I would be fine if struct mmk_stub was renamed to struct mmk_stub_ however.
There was a problem hiding this comment.
It does break API, but I could only find one place where the mmk_stub alias was being used instead of struct mmk_stub *. I figured I'd normalize to the struct <name> syntax instead.
There was a problem hiding this comment.
This is mostly about API usage -- mmk_stub is an opaque type, so having the user type struct mmk_stub in C is just misleading because you'd expect to be able to change or read some fields in there.
In other words, I am fine with keeping struct and union keywords when it's possible to access the underlying type, and this is not one of these situations.
| off[j++] = i; | ||
| if ((s->n_type & N_EXT) && ((s->n_type & N_TYPE) == N_UNDF)) { | ||
| const char *symname = strtab + s->n_un.n_strx; | ||
| if (!strcmp(symname + 1, name)) off[j++] = i; |
There was a problem hiding this comment.
nit: I'd prefer if the if statement was over two lines.
| if (lc->cmd == MMK_LC_SEGMENT) { | ||
| const segment_cmd * sc = (void *) lc; | ||
| if (strstr(sc->segname, "TEXT") != NULL) { | ||
| text = sc; | ||
| } else if (strstr(sc->segname, "LINKEDIT") != NULL) { | ||
| linkedit = sc; | ||
| } | ||
| } |
| # define MMK_COMMA_APPEND(_, prefix, content) (void) (prefix content), | ||
|
|
||
| # define mmk_struct_initialize(variable, ...) \ | ||
| (MMK_EXPAND(MMK_APPLY_N_(MMK_COMMA_APPEND, MMK_VA_NARGS(__VA_ARGS__), variable, __VA_ARGS__)) \ |
There was a problem hiding this comment.
Hah, what a clever hack -- I love it!
| return dst; | ||
| } | ||
|
|
||
| va_list & mmk_assign(va_list & dst, va_list src) { |
|
This grew massive and hard to review. I'll close this now, and open smaller PRs with these and other fixes. Hope you can take a look @Snaipe. |
Signed-off-by: Stephen Brawner <brawner@gmail.com>
Precisely what the title says. This patch allows
Mimickto be used in C++ tests for C API.Depends on #11 and #12. Only 31c6ce0, 8381645, and 0b67235 are specific to this patch.