Re: Cygwin Package ncurses 6.3-20221224 ABI > 9 config patches

[Brian Inglis]

On 2023-01-01 10:18, Sam James wrote:
> On 1 Jan 2023, at 15:38, Thomas Dickey wrote:
> > On Sun, Jan 01, 2023 at 08:19:36AM -0700, Brian Inglis wrote:
> > > When building ncurses 6.3-20221224, got compile errors with
> > > NCURSES_MOUSE_VERSION not set in curses.h, coming from configure.in,
> > > autoconf generated as configure, run with option --with-abi-version=10,
> > > which has an issue with the case $cf_cv_abi_version block setting only for
> > > ABI 5-9.
> > > I fixed my issue by setting the variable in a case branch handling 1[0-9],
> > > initially to 4, but that is checked and reset to '', so set to 3, although
> > > it should probably be 2 as the ABI version is originally 6.

> > ah... I overlooked that some packagers renumber the ABI's.
> > The basic problem there is that someone notices a functional change
> > which doesn't affect the binary interface, and decides to increment
> > the ABI.  (I don't recall anyone asking me before they did that).
> > I suppose a fix here would be to add options to set the mouse-version
> > and perhaps a few other ABI-dependent features explicitly (and warn rather
> > than error-out when a mismatch is seen).

> I'm very surprised any packagers do that. Definite misbehaviour. :(

The config option is there for historical reasons, and our previous maintainer 
had it at 8 by 5.6 (15 years ago in 2008), and 10 for 5.9 (10 years ago in 
2013), so we could revert, but every interactive shell, editor, terminal 
multiplexor, and all packages using readline or ncurses, would have to be 
rebuilt and retested (some probably now unmaintained), and I would not be 
popular *trying* to make that call, nor happy when such an upgraded package was 
abruptly withdrawn.

--
Take care. Thanks, Brian Inglis                 Calgary, Alberta, Canada

La perfection est atteinte                      Perfection is achieved
non pas lorsqu'il n'y a plus rien à ajouter     not when there is no more to add
mais lorsqu'il n'y a plus rien à retirer        but when there is no more to cut
                        -- Antoine de Saint-Exupéry

[BTDT! I recently updated Cygwin grep with stderr warnings deprecating e/fgrep 
commands, had it in test status for a month, with requests for feedback to all 
maintainers and developers when announced prior to test availability, at test 
availability, and a month later (in theory if they read the list), and promoted 
to current stable another month later, with another obvious note in the release 
announcement to all users.
Only when it was then upgraded on their systems, did the maintainers, 
developers, and project leads complain and request reversion of those messages 
in a new release! ;^> ]