[tin-bugs] (fwd) Cross-compilation of tin

[Thomas Dickey]

On Sun, Feb 05, 2023 at 06:03:19AM +0100, Urs Janßen wrote:
> Hello Thomas,
> 
> would you mind to answer the following mail as it is mainly your code
> and macro. IIRC free(NULL) was an issue > ~25 years ago on several
> operating systems as it caused crashes (but I can't remeber which ones).

I encountered this with SunOS 4 (and probably HP-UX) in getting the sam editor
to work (perhaps 1993).

fwiw, SunOS 4 was in use for development after 2000 :-)
 
> And at least it should include "bool.h" not <bool.h> and

The <bool.h> should work, assuming that the compiler's -I options are
set up properly (to point to tin's include-directory).

> <stdlib.h> and <string.h> not unconditionally.

agree - those should be ifdef'd.  For anything produced over the
past 20 or so years, only unistd.h has been a problem, but
recalling the xpg3 fixups that I did a couple of years ago,
the conditional inclusion still applies.
 
> Thanks,
> urs

> Date: Thu, 2 Feb 2023 22:36:59 +0300
> From: uis <uis9936 at gmail.com>
> To: tin-bugs at tin.org
> Subject: [tin-bugs] Cross-compilation of tin
> 
> While trying to cross-compile tin using crossdev on Gentoo, it failed to do
> because makecfg.c includes tin.h which includes gsasl.h. Usually it is not
> unexpected to do so, given the fact that sasl USE-flag was set, but it was
> compiling tool for CBUILD. So gsasl was required on build system, which is
> very wierd.

...well, a fix for that might be helpful.

> Insted of simply installing tin into build system, I looked at source and
> found that only thing needed from tin.h is FreeIfNeeded macro. And this
> macro is free with condition of pointer being non-null. This macro doesn't
> have different bahaviour compared to simple free(), if you look man 3 free,

sure - tin builds/works with old systems that some people don't care about.

> it says "If ptr is NULL, no operation is performed".

reading the manual page may help (see the section on conformance).

> I changed includes and code and now it should work. Patch attached

the suggested patch isn't relevant to the reported bug.

> --- src/makecfg.c	2021-12-22 16:24:50.000000000 +0300
> +++ src/makecfg.c	2023-02-02 22:34:15.804665881 +0300
> @@ -38,11 +38,11 @@
>   */
>  
>  
> -#define __BUILD__
> -#define PROTO_H 1	/* don't include proto.h */
> -#ifndef TIN_H
> -#	include "tin.h"
> -#endif /* !TIN_H */
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include <string.h>
> +#include <bool.h>
> +#include <ctype.h>
>  
>  #define L_CURL '{'
>  #define R_CURL '}'
> @@ -542,8 +542,8 @@
>  	m = all_data;
>  	while (m) {
>  		n = m->link;
> -		FreeIfNeeded(m->name);
> -		FreeIfNeeded(m->type);
> +		free(m->name);
> +		free(m->type);
>  		free(m);
>  		m = n;
>  	}

> _______________________________________________
> tin-bugs mailing list
> tin-bugs at tin.org
> http://lists.tin.org/cgi-bin/mailman/listinfo/tin-bugs



-- 
Thomas E. Dickey <dickey at invisible-island.net>
https://invisible-island.net
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <http://lists.tin.org/pipermail/tin-bugs/attachments/20230205/2c63dee0/attachment.sig>