ct.c tuneups from NCBI as of May 2024

[ucko]

This PR contains two tuneups to ct.c from #555 that git-deps flagged as interdependent, though it looks like that may have been a false positive.

[freddy77]

Merging first commit.

About the second, I'm doing some tests. Looks like the check on Sybase library is done while calling ct_param, not sending. Also the error is pretty different, I'm getting (CS_CHAR_TYPE):

Open Client Message: 0x2f658890 0x2f674180
number 0x101018a layer 1 origin 1 severity 1 number 138
msgstring: ct_param(): user api layer: external error: A data length of 25600 exceeds the maximum length allowed for CHAR data.
osstring: (null)
ct_param(char) failed

Also it looks like data for parameters are copied during ct_param (changing them after do not affect the data used in the query).

[ucko]

Thanks! I'll take a closer look at the second.

[ucko]

Ah, yeah, looks like I went with an existing error code rather than checking what the SAP implementation does.

[freddy77]

No rush. But the difference about ct_param getting the data is more concerning to me (well, not that affect this PR).
Consider a program that:

• computes value allocating memory;
• pass value to ct_param;
• free the value;
• calls ct_send.
  At the time of ct_send if we attempt to get the data having just saved the pointer value could be garbage or released to OS, so potentially causing corruption (invalid data inserted) or crashes.

Anyway, OT.