FreeTDS Development Group

["Bill Thompson" <thompbil AT exchange.uk.ml.com>]

...although please feel free to chip in if you have a view...

I'm wanting to implement some of the ct-library option flags that can be
manipulated with ct_config(), a function that is currently unimplemented.

In order to do this, I'm going to need somewhere to store them. I thought
I'd ask your opinion, as I'm not *really* an expert in the ct-library
code.

ct_config() is defined as follows:

CS_RETCODE ct_config(CS_CONTEXT *ctx, CS_INT action, CS_INT property,
CS_VOID *buffer, CS_INT buflen, CS_INT *outlen)

given this, it seemd logical to me to store these flags somewhere in the
CS_CONTEXT structure. This is currenty defined as follows:

typedef struct cs_context
{
    CS_INT date_convert_fmt;
    int (*_clientmsg_cb)(void*, void*, void*);
    int (*_servermsg_cb)(void*, void*, void*);
    TDSCONTEXT *tds_ctx;
} CS_CONTEXT;

As these flags are for ct-library only. It made sense to me to store them
in the CS_CONTEXT structure itself, rather than the TDSCONTEXT structure
therein.

So, what I propose to do is as follows:

typedef struct cs_config
{
    short cs_expose_formats;
} CS_CONFIG;

typedef struct cs_context
{
    CS_INT date_convert_fmt;
    int (*_clientmsg_cb)(void*, void*, void*);
    int (*_servermsg_cb)(void*, void*, void*);
    TDSCONTEXT *tds_ctx;
    CS_CONFIG config;
} CS_CONTEXT;

where "cs_expose_formats" is the particular flag I am interested in
implementing at the moment. More would be added to the CS_CONFIG structure
as and when we support them.

I'd like to get your opinion on whether I've done the right thing.

Bill