Swap.GetEndian() breaks code

[mikanystrom]

Hi everyone,

This change...

scc406110> pwd
/nfs/site/disks/zsc9_fwr_lib_char_008/mnystroe/cm3-fork/m3-libs/m3core/src/swap
scc406110> git blame Swap.i3
cfac2c7 m3-libs/libm3/src/uid/Common/Swap.i3 (Olaf Wagner 2001-01-24 12:24:49 +0000 1) (* Copyright (C) 1989, Digital Equipment Corporation *)
...
cfac2c7 m3-libs/libm3/src/uid/Common/Swap.i3 (Olaf Wagner 2001-01-24 12:24:49 +0000 12) TYPE Endian = {Big, Little};
cfac2c7 m3-libs/libm3/src/uid/Common/Swap.i3 (Olaf Wagner 2001-01-24 12:24:49 +0000 13)
50f54fc m3-libs/m3core/src/swap/Swap.i3 (Jay Krell 2022-03-07 03:30:45 -0800 14) PROCEDURE GetEndian(): Endian;
cfac2c7 m3-libs/libm3/src/uid/Common/Swap.i3 (Olaf Wagner 2001-01-24 12:24:49 +0000 15)

breaks existing code.

Yes, I realize that the old way, the global variable "endian", is not ideal.

But the standard installation of Modula-3 at Intel has it, and there's code around that uses it. (Specifically, the rdwr library, but maybe others too.)

The interface also moved...I think that's less of a problem since most programs will pull in libm3 and m3core and at least can be made to work with either version. But deleting "endian" is a problem.

What problem did this solve?

The implementation of GetEndian is

PROCEDURE GetEndian(): Endian =
BEGIN
RETURN endian;
END GetEndian;

so "endian" is still there.

Would anyone mind if I change it so that "endian" is back to being visible in the interface, but marked <OBSOLETE>?

[VictorMiasnikov]

Why not?