Re: QtmsCreateSendEmail

["Peter Clifford" <pete@xxxxxxxxxxxxxxxxxxxxx>]

I started off with the attachments as a character field but changed it to a 
pointer (admittedly, it should have been by VALUE, not CONST, I agree) 
because the documentation confusingly refers to the 6th and 8th parameters 
needing to be "null". If I define this as character, I'm not sure it's 
possible to omit these parameters and still use the error structure. Or 
would *OMIT be allowable here?

Pete


"Scott Klement" <midrange-l@xxxxxxxxxxxxxxxx> wrote in message 
news:mailman.2946.1224400035.30674.midrange-l@xxxxxxxxxxxxxxx
> Hi Peter,
>
> Sorry about this, but I'm going to pick your prototype apart ;)
>
> > D CrtSndMail      PR                  EXTPROC('QtmsCreateSendEmail')
> > D   iaRecipients             32767A   CONST OPTIONS(*VARSIZE)
>
> You'll have problems if iaRecipients ever exceeds 32767 bytes, since
> you've defined this as CONST, which gives RPG the right to make a copy
> of it into temporary memory, and pass the copy.  If it makes a copy,
> it'll only pass the first 32767 bytes...    So if there's any chance of
> exceeding that size, remove the CONST.
>
>
> > D   iuRcvCount                  10U 0 CONST
>
> This should be 10I 0, not 10U 0.  That's not the cause of your problems,
> but the API does designate this as a signed field, not unsigned, so
> you're best off using 10I 0.
>
>
> > D   iaRcvFormat                  8A   CONST
> > D   iaNote                   32767A   CONST OPTIONS(*VARSIZE)
> > D   iaNoteFormat                 8A   CONST
> > D   ipAttachments...
> > D                                 *   CONST OPTIONS(*NOPASS)
>
> You have ipAttachments this defined as a pointer -- but the API is
> expecting character.  That's simply incorrect, and will produce the
> dreaded "unexpected results".
>
> You may be thinking of passing a pointer by VALUE.  You see, passing a
> normal field by reference causes the address of the field to be passed
> (under the covers).  Therefore, passing a pointer by value is equivalent
> to passing a regular field by reference.
>
> However, in this case, you've passed a pointer by reference -- which is
> in no way equivalent to passing a character field.
>
>
>
> > D   iuAtcCount                  10U 0 CONST OPTIONS(*NOPASS)
>
> Again, should be 10I 0.
>
>
> > D   ipAtcFormat                   *   CONST OPTIONS(*NOPASS)
>
> Again, this should be a character field, not a pointer.
>
>
>
> > EVAL apierror:x
> >    00000     00000110 00000014 E3C3D7F5 F3F0F8F0   - ........TCP53080
> >    00010     A2A49797 40404040 40404040 40404040   - supp
>
> It seems strange that debug is only showing the first 32 bytes of a
> sructure that you claim is 272 bytes long?!   Or did you abbreviate it
> when you posted it to the mailing list?
>
> At any rate, the crucial problems I see are with the two parameters where 
> you have them defined as pointers instead of character.