[Cobolforgcc-devel] Common COBOL library
Status: Pre-Alpha
Brought to you by:
timjosling
From: Bernard G. <bg...@fr...> - 2002-09-28 07:05:14
|
Hi all, Good news and bad ones! On the good news front, I'm very pleased of the final ratification of COBOL 2002 by the ISO. On the bad ones, we (Opensource Cobol compiler projects developers) have still a long way to go, just to implement the 85/89 standard. I think it is about time to try to work a little more together, following the tracks of David Essex and Keisuke Nishida. So many things need still to be done (not in any order), like : - other file support systems (Inodb, RMS, ...) - decent sort library and interface - Sql preprocessor for ODBC and MySQL - COBOL prettyprinter (like indent for C) - integration into GDB, DDD and IDE's ... So I propose the following (I think Tim did agree on the principle): design a common libcob, and if necessary, migrate the corresponding compilers. Just to start the discussion, I see two main approaches: either take an existing one, "donate" it to the cobol project, and upgrade to the specs, or just start from scratch again. No surprise, I'm in favour of the first; but which one? Well, it might depend on the specs we adopt. So I think the best start is the specs for: - the way the cobol field is accessed from the routine - a descriptor for a cobol field - a descriptor for a cobol file Just to start the ball rolling, I will send this message to the compilers mailing list, but I think the discussion should be carried on in the cobol-util ML. First of all, Keisuke, what was the reason behind having a descriptor with two pointers, one for the attr desc and the other for the data, except economy of paramters for the routines and possible factorisation of field_desc ? Hoping to concentrate forces ... Bernard Giroud |