Debian bug report logs - #1645
termcap

Package: ncurses-developer; Reported by: Ian Jackson <iwj10@cus.cam.ac.uk>; 20 days old.

Message received at debian-bugs:

From cus.cam.ac.uk!iwj10 Fri Oct 13 06:03:18 1995
Return-Path: <iwj10@cus.cam.ac.uk>
Received: from pixar.com by mongo.pixar.com with smtp
	(Smail3.1.28.1 #15) id m0t3jld-0002G3C; Fri, 13 Oct 95 06:03 PDT
Received: from bootes.cus.cam.ac.uk by pixar.com with SMTP id AA01087
  (5.67b/IDA-1.5 for debian-bugs-pipe@mongo.pixar.com); Fri, 13 Oct 1995 06:01:18 -0700
Received: by bootes.cus.cam.ac.uk 
	(Smail-3.1.29.0 #36) id m0t3jfx-000C0ZC; Fri, 13 Oct 95 13:57 BST
Received: by chiark
	id <m0t3j7i-0002Z8Z@chiark.al.cl.cam.ac.uk>
	(Debian /\oo/\ Smail3.1.29.1 #29.33); Fri, 13 Oct 95 13:22 BST
Message-Id: <m0t3j7i-0002Z8Z@chiark.al.cl.cam.ac.uk>
Date: Fri, 13 Oct 95 13:22 BST
From: Ian Jackson <iwj10@cus.cam.ac.uk>
To: "Eric S. Raymond" <esr@locke.ccil.org>
Cc: debian-bugs@Pixar.com, "Zeyd M. Ben-Halim" <zmbenhal@netcom.com>
Subject: Re: Bug#1645: termcap 
In-Reply-To: <Pine.3.89.9510122322.A751-0100000@locke.ccil.org>
References: <m0t3Yee-0002ZTZ@chiark.al.cl.cam.ac.uk>
	<Pine.3.89.9510122322.A751-0100000@locke.ccil.org>

Eric S. Raymond writes ("Re: Bug#1645: termcap "):
> Hmmm.  Ctrl-Z problem?  Please explain.  We had a bug involving ^Z in some 
> older versions, but I think it's fixed now.

dselect, our package selection program, does (basically) the following:

 initscr();
 ... lots of (n)curses stuff, user makes menu selections, &c ...
 ... user selects menu option to run some program ...
 clear(); refresh(); endwin();
 ignore SIGQUIT, SIGINT, SIGHUP  (sigaction(...,{SIG_IGN},&save))
 fork/exec a subprocess
 waitpid(pid,&status,0);
 restore SIGQUIT, SIGINT, SIGHUP  (sigaction(...,&save,))
 sleep(1)
 initscr();
 ... go back into ncurses ...

If you hit ^Z while the subprocess is running (the subprocess takes
around a second) it seems to work only some of the time if at all.
Often ^Z has no effect at all; other times it works once but not the
second time you try it; sometimes it appears to get you back into
dselect somehow and leaves (presumably) a stuck subprocess around.

What I think it ought to do is background the whole assembly of
dselect and its subprocess.

(Come to think of it, I don't think I should be ignoring SIGHUP.  I'll
remove it from the list of signals-to-be-ignored.)

> I'm not sure I have the attributions straight here.  If you'll send me the 
> conXY entries, I'll put them in the terminfo master.

I think that would probably be a bad idea.  They're obsolete,
Linux-specific and generally very horrible.  Bruce ?

> OK, I wasn't clear on that.  Bruce, you guys have good karma with me (and
> I expect with Zeyd, too) because you've provided quite a few high-quality
> bug reports (precise, replicable, etc) that we've been able to use to
> improve the package.  If there is anything we can do in the way of package
> reorganizing, more or different makefile scaffolding, documentation, or
> whatever, to make the Debian build process cleaner, please don't be shy
> about making suggestions. 

Speaking personally as the main person using ncurses directly in a
specifically-for-Debian program, the best thing you can do for me is
to fix bugs :-).

> I have one question that may relate to this.  A lot of the older terminfo
> entries in the latter half of the master file (after the cut mark) are
> just junk from the P.O.V, of linux users (who's gonna hook up a netronics
> or an exidy 2500 to a linux box at this late date?).  They take up what
> might be considered a substantial amount of space on smaller systems 
> (the whole tree is 1293 blocks on my machine and I'm sure somewhere between 
> half and three-quarters of that is old entries). 
> 
> I've been thinking about doing something about this but am not sure it's
> worth the effort.  I'm not clear on whether the disk space savings from being
> selective are worth the slight additional installation complexity.
> 
> Would it be a net win for you if we:
> 
> 1. Provided a configuration option to only install the newer entries in the
>    first half of the file?
> 
> 2. Provided some interactive Perl thingy for culling out the terminfo tree
>    after compilation?
> 
> 3. Did something else clever at the cost of giving you more 
>    default-configuration decisions to make: (fill in the blank here)

Bruce, is it worth us putting really obscure terminals into a separate
.deb package ?

I've been thinking that `ncurses-runtime' is a bad package name.  I
think `terminfo-data' would be a better one.  We could have
`terminfo-base' for the minimal terminfo on the base disks, and then
if you wanted a separate package you could have `terminfo-obscure'.

Ian.

Acknowledgement sent to Ian Jackson <iwj10@cus.cam.ac.uk>:
Extra info received and forwarded. Full text available.