Zsh Mailing List Archive Messages sorted by: Reverse Date, Date, Thread, Author

Re: Make Failure on SunOS-4.1

X-seq: zsh-workers 14095
From: Clint Adams <schizo@xxxxxxxxxx>
To: Bart Schaefer <schaefer@xxxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Make Failure on SunOS-4.1
Date: Wed, 25 Apr 2001 10:57:30 -0400
Cc: Peter Stephenson <pws@xxxxxxx>, Zsh hackers list <zsh-workers@xxxxxxxxxx>
In-reply-to: <1010425144152.ZM11080@xxxxxxxxxxxxxxxxxxxxxxx>; from schaefer@xxxxxxxxxxxxxxxxxxxxxxx on Wed, Apr 25, 2001 at 02:41:52PM +0000
Mailing-list: contact zsh-workers-help@xxxxxxxxxx; run by ezmlm
References: <Tc0a88d015321834e70@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx> <1010425144152.ZM11080@xxxxxxxxxxxxxxxxxxxxxxx>

> We could back out the `termcap' parameter entirely, and we could back out
> the ability to use ${(k)terminfo} -- those were the changes that started
> requiring the ncurses header files to be included even on systems that
> have the old termcap as well.

Or, since I think that the only dependency on term.h is for the
{bool,num,str}codes, those could be made static under whatever
circumstances would call for that.
 
> Or we could change the termcap-search-order in configure.in to always try
> to find ncurses first, and skip defining HAVE_TERMCAP_H when we find the
> ncurses headers.  (My understanding is that in the distant past of zsh,
> there were some who objected to linking the shell against curses when it
> would be sufficient to link against termcap, because curses is bloated.)
>
> Or we could add another --enable flag, which switches between the two
> states described above (in which case it should probably disable the
> terminfo parameter entirely rather than just crippling it).

All of these seem reasonable to me.

References:
- Re: Make Failure on SunOS-4.1
  - From: Peter Stephenson
- Re: Make Failure on SunOS-4.1
  - From: Bart Schaefer

Messages sorted by: Reverse Date, Date, Thread, Author