Comment author: administrator

The second form of union-decl (according to p. 2.5 of camlidl
manual) does not work:

[union.idl]
typedef union switch(int ctx)
{
case 1: int intVal;
case 2: char charVal;
} UNION;

As shown in the manual, each "case" label must syntactically be an
identifier, e.g.

typedef union switch(int ctx)
{
case FIRSTCASE: int intVal;
case SECONDCASE: char charVal;
} UNION;

Otherwise, the IDL->Caml mapping cannot guess the Caml constructor
names for each case of the union type.

• Xavier Leroy

Comment author: administrator

Behavior conforms to spec.

Comment author: administrator

Xavier Leroy Xavier.Leroy@inria.fr writes:

The second form of union-decl (according to p. 2.5 of camlidl
manual) does not work:

[union.idl]
typedef union switch(int ctx)
{
case 1: int intVal;
case 2: char charVal;
} UNION;

As shown in the manual, each "case" label must syntactically be an
identifier, e.g.

typedef union switch(int ctx)
{
case FIRSTCASE: int intVal;
case SECONDCASE: char charVal;
} UNION;

Otherwise, the IDL->Caml mapping cannot guess the Caml constructor
names for each case of the union type.

You are right, sorry. I've tried to find out why the following construct
from wtypes.idl generates syntax error (you know, camlidl in fact does not
point to the correct place of error):

typedef union _userCLIPFORMAT switch(long fContext) u
{
case WDT_INPROC_CALL: DWORD dwValue;
case WDT_REMOTE_CALL: [string] wchar_t * pwszName;
} userCLIPFORMAT;

and have simplified it in the wrong place. The real problem was that "u"
after switch (don't know what it means for MIDL). Sorry once again.

Hope to hear from you soon,
Dmitry