[pacman-dev] Pacman doesn't handle properly the multiple satisfiers, localdb corruption

[Xavier]

On Mon, Jul 16, 2007 at 07:25:50PM +0200, ngaba at petra.hos.u-szeged.hu wrote:
>> Yes indeed, it might be safer to remove.
> OK, my patch attached, please test it.
> Bye, ngaba
>
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>

> diff --git a/lib/libalpm/package.c b/lib/libalpm/package.c
> index a18dc1f..5c94141 100644
> --- a/lib/libalpm/package.c
> +++ b/lib/libalpm/package.c
> @@ -844,6 +844,9 @@ int _alpm_pkg_splitname(const char *target, char *name, char *version, int witha
>  
>  /* scan the local db to fill in requiredby field of package,
>   * used when we want to install or add a package */
> +
> +/* A depends on B through n depends <=> A listed in B's requiredby n times */   
> +/* n == 0 or 1 in almost all cases */ 

I failed to see why we would require to put the same package more than once
in requiredby.
I think it works that way, but it isn't necessary, is it?
At least I couldn't find a pactest for a case where this would be needed. Is
there one?

Actually, I've nothing against these multiple requiredby, but I think Dan
doesn't like it. I believe it's mostly because of that bug where requiredby
were added wrongly multiple times.
While in this case, having them multiple times would actually have a meaning
(multiple depends), and anyway like you said, it's a corner case that would
probably never happen in practice :)

So I actually like this patch in its current state a lot (as well as the
other one for add_commit).
But I think these may still need some review/thoughts.