Re: mariadb106-client failure to run mysql

[PHO <pho%cielonegro.org@localhost>]

On 1/15/22 6:50 PM, Jaromír Doleček wrote:
> I believe the +BUILDLINK_TRANSFORM+= rm:-ltermcap should go into
> ncursesw buildlink - after all, that is the one which provides it's
> own termcap things.

Probably. It should remove -lterminfo too. Also 
devel/ncurses/buildlink3.mk should do the same, as ncurses and ncursesw 
both provide termcap and terminfo interfaces.

> Of course, ncursesw has no business providing it's own termcap things,
> and that should be the root problem to fix.

I think it is too harsh to say that, because the only reason ncurses has 
a termcap interface is to help applications which still use termcap. Its 
termcap implementation is in fact just an emulation layer built on top 
of terminfo. See:
https://invisible-island.net/ncurses/man/curs_termcap.3x.html

> Le sam. 15 janv. 2022 à 06:36, PHO <pho%cielonegro.org@localhost> a écrit :
> > On 1/15/22 7:50 AM, Chavdar Ivanov wrote:
> > > Just a ping - has anybody seen that? I see mariadb106 is now the
> > > default mysql version and can't imagine nobody were using ncursesw. I
> > > don't understand why it gets linked with libncursesw - the generated
> > > CMakeCache.txt references only libncurses.
> > >
> > > If I take out 'CURSES_DEFAULT=ncursesw' then the executable starts as expected.
> > >
> > > On Wed, 5 Jan 2022 at 13:02, Chavdar Ivanov <ci4ic4%gmail.com@localhost> wrote:
> > > > Hi,
> > > >
> > > > I have CURSES_DEFAULT=ncursesw in my /etc/mk.conf. On two separate
> > > > systems I am getting the same crash, immediately after starting mysql
> > > > from the latest mariadb106 package:
> >
> > I haven't tried mariadb106-client, but presumably the same issue has
> > happened on audio/ncmpcpp. I have applied a workaround with some
> > uncertainties remained:
> > http://mail-index.netbsd.org/pkgsrc-changes/2021/11/29/msg244229.html