Provide a system-level iconv table for SYSVAL(QCCSID) to CCSID(1208) and CCSID(1200) Cached

As part of our ongoing efforts to optimize character encoding conversion in C and C++ applications on IBM i, we are currently utilizing a multi-step process to handle conversions between the JOB CCSID (an EBCDIC CCSID) and UTF-16 (CCSID 1200). Specifically, we convert from the JOB CCSID to CCSID 1208 (UTF-8) and then from 1208 to 1200 (UTF-16), and for reverse conversions, we perform a similar two-step process.

While this approach has been optimized to some extent, it is not always practical, especially in environments where performance and memory efficiency are critical. Implementing this conversion process for each program and scope often results in redundancy and unnecessary resource consumption, particularly when dealing with repetitive conversions between the JOB CCSID and the standard UTF-8 (1208) and UTF-16 (1200) encodings.

Therefore, I am requesting that IBM provide an optimized, system-level solution for handling these conversions. Specifically, I propose the following enhancement:

1. Generation of Pre-Configured Iconv Tables at IPL (or another suitable time):

At IPL time (or at another appropriate time), IBM could generate and cache the equivalent of the iconv conversion tables that handle the following:

• QCCSID system value CCSID to CCSID 1208 (UTF-8)

• CCSID 1208 (UTF-8) to CCSID 1200 (UTF-16)

• Reverse conversions for both of the above.

2. System Access to these Pre-Configured Conversion Tables:

Developers could then access these pre-configured conversion tables via a special or symbolic interface, similar to how CCSIDs are handled in SQL and RPG. For example:

iconv(*JOBTO1208...) 
iconv(*1208to1200...)
iconv(*1200to1208...)
iconv(*1208toJOB...) 

This would eliminate the need for developers to construct these tables at runtime, saving both time and system resources.

3. Efficient and Redundant-Free Conversion:

While I don't mind doing the 2-step conversions in my code, by providing these system-level conversion tables, developers would only need to access the pre-configured conversion resources rather than creating and managing them in every individual program and scope. This would streamline the conversion process, reduce unnecessary duplication, and improve both performance and maintainability, not to mention make it so much easier to implement this kind of support in applications.

I believe this enhancement would align with the integrated conversion capabilities already present in SQL and RPG IV, where specifying a CCSID on a variable automatically triggers the appropriate conversion. In contrast, C and C++ developers currently lack a similar, optimized system-level solution, resulting in inefficient and often redundant conversions.