for that matter

Hello, All then the indicated methods imply “interaction” of the program loaded and parser running in ROM. look at http://zxdocs.fatal.ru/coding/module.zip - it’s already very well solved there this question, in principle with minor modifications the same method can be considered optimal. In general, my deep conviction is that it is stupid to implement parsing call during the operation of the program itself - when loading the program it is still A LOT of CPU time and preprocessing of the program body are spent will increase this time slightly (loading one sector of 256 bytes takes about 20-30 thousand clock cycles, and the average utility program takes 5-6 sectors, many times more toys and demos; process for setting up direct calls Routine from ROM consumes qualitatively fewer cycles - most likely it will fit into one interrupt), however, the absence of a parser during program EXECUTION will lead to simply to qualitatively speed up the OPERATION of the code. For example, take programs for multiplying two one-byte numbers. The program itself takes about 500 cycles. And the parser is 200. It turns out that from all the time execution spends about 40% of the time on determining the address of the call???? And This is every challenge!!! That is why all branded toys were used either directly ROM calls (directly Call ) or even your own programs, which we know exactly where they are.