Re: patches to make it compile with very old compilers

On Jul 17, 2009, at 11:39, "Bernhard R. Link" <[email protected]> wrote:

> Here are two patches to make ratpoison compile with older
> compilers.
>
> The first one fixes an actual bug in putting libraries in library
> flags instead of libraries:
>
> Subject: [PATCH] move libraries from LDFLAGS to LDADD to avoid  
> problems in argument order
>
> diff --git a/src/Makefile.am b/src/Makefile.am
> index 715c47b..b4ce142 100644
> --- a/src/Makefile.am
> +++ b/src/Makefile.am
> @@ -22,7 +22,8 @@
> bin_PROGRAMS        = ratpoison
> MAINTAINERCLEANFILES = Makefile.in config.h.in
>
> -AM_LDFLAGS=${X_LDFLAGS} ${X_LIBS} ${X_EXTRA_LIBS} ${XFT_LIBS} $ 
> {HISTORY_LIBS}
> +AM_LDFLAGS=${X_LDFLAGS}
> +ratpoison_LDADD=${XFT_LIBS} ${X_LIBS} ${X_EXTRA_LIBS} ${HISTORY_LIBS}
> AM_CPPFLAGS=${X_CFLAGS} ${XFT_CFLAGS}
>
> ratpoison_SOURCES    = actions.c        \
>
> I've applied this as "obvious trivial bugfix".
>
> The second patch[1] stops using some c99 features to make it compile  
> with
> gcc version 2, which I am very unsure about:
>
> --- a/src/history.c
> +++ b/src/history.c
> @@ -65,7 +65,7 @@ extract_shell_part (const char *p)
>
> struct history_item {
>    struct list_head node;
> -    char line[];
> +    char line[1];
> };
>
> static struct history {
> @@ -155,7 +155,7 @@ history_add_upto (int history_id, const char
> *item, size_t max)
>      return;
>
>   item_len = strlen(item);
> -  i = xmalloc (sizeof(struct history_item) + item_len + 1);
> +  i = xmalloc (sizeof(struct history_item) + item_len);
>
>   memcpy (i->line, item, item_len + 1);
>   list_add_tail (&i->node, &h->head);
>
> Pros:
> - makes it work with some gcc 2.9x compilers.
> (I do not know how many, as linkedlist.h also uses enhancements,
> so that even with this it is not c89 compatible but at most some
> older variants of gnu89).

IIRC the Linux linkedlist includes more portable non-gcc-specific  
idioms in case of an old GCC. You might try bringing those into the RP  
linked list.

>
> Cons:
> - we are 10 years after c99
> - might confuse any array length checkers (though they might have
>  an exception for [1], as that does not make much sense otherwise).
>
> What do you think?
>
> Hochachtungsvoll,
>    Bernhard R. Link
>
> [1] written and tested by Ineiev <[email protected]>
>
>
> _______________________________________________
> Ratpoison-devel mailing list
> [email protected]
> http://lists.nongnu.org/mailman/listinfo/ratpoison-devel