Hi Patrick
What language are you using?
In C++ anyway you have to set the value, as well as set the flag indicating
that it is indeed
set see __isset in the generated code.
Dont know if thats the cleanest thing, but in my local headers I have the
following macros
#define *THRIFT_HAVE_OPT_FIELD*(O,M) \
((O).__isset.M)
#define *THRIFT_MARK_OPT_FIELD*(O,M) \
(O).__isset.M = true
#define *THRIFT_SET_OPT_FIELD*(O,M,V) { \
(O).M = (V); \
*THRIFT_MARK_OPT_FIELD*(O,M); \
}
And would use it as follows
*THRIFT_SET_OPT_FIELD*(theSession,btest , true);
--
+---------------------------------------------+-------------------------+
| Nevo Hed | +1-617-302-6175 |
| Senior Principal Embedded Software Engineer | nhed at bamboom dot com |
| Bamboom Labs | http://www.bamboom.com |
+---------------------------------------------+-------------------------+
On Mon, Aug 8, 2011 at 4:51 AM, Patrick Nick <[email protected]> wrote:
> Hi all,
>
> I have started experimenting with Thrift. After running the provided
> tutorial in Java I decided to extend it with my own datatype.
>
> struct Session {
> 1: bool isSessionAlive,
> 2: string sessionID,
> 3: optional string creatingIP,
> 4: optional bool btest,
> }
>
> However, I am observing strange behaviour of the booleans. The server
> creates a session object and fills all 4 fields, setting both the mandatory
> and the optional bool to true.
> When it is sent to the client I read out the values, and at that point the
> mandatory bool "isSessionAlive" is true, but the optional bool "btest"
> evaluates to FALSE.
> If I change the definition file and take away the optional flag, making
> "btest" mandatory, it evaluates to true as expected. The rest of the code
> remains unchanged.
>
> It seems as if I can not get optional booleans to evaluate to true after
> sending an object. Is this a functionality which I don't understand or a
> bug?
> I have not tried in other languages so far, only in Java.
>
> Kind regards,
> Patrick
>
