Re: Bugs in To: entry

[Emmanuel <e allaud wanadoo fr>]

On 2001.10.11 14:07 Emmanuel wrote:
> 	Hi all,
> here is a little bug but still annoying. When you compose a new message,
> type something in the to: address field, select it and erase it with
> backspace or del, as you want, then open your address book, select one
> person, you'll see the correct address in the to: field. But now press
> send
> button, this will erase the To: field, causing the message not to be sent
> (because Balsa does not send message with empty address).
> It's related with libbalsa_..._clear_to_send function, but I was not able
> to really find it. So if someone who knows this code better than me.
> Moreover look at this function in libbalsa/address-entry.c :
> 
> /*************************************************************
>  * libbalsa_address_entry_show:
>  *     Shows the widget.  This will work out the aliases,
>  *     and set the widget text to that, and then call
>  *     libbalsa_address_entry_draw_text() to draw the
>  *     widget.
>  *
>  *   arguments:
>  *     address_entry: the widget.
>  *
>  *   results:
>  *     modifies address_entry
>  *
>  *   FIXME:
>  *     - Adding the lists together should be easier.
>  *     - It should make more use of helper functions
>  *       that got written after this function got written,
>  *       like libbalsa_make_address_string()
>  *************************************************************/
> void
> libbalsa_address_entry_show(LibBalsaAddressEntry *address_entry)
> {
>     GtkEditable *editable;
>     GString *show;
>     GList *list;
>     emailData *addy;
>     gchar *out;
>     gint cursor, start, end;
>     gboolean found;
>     inputData *input;
>     gint tmp_pos;
> 
>     g_return_if_fail(address_entry != NULL);
>     g_return_if_fail(LIBBALSA_IS_ADDRESS_ENTRY(address_entry));
>     g_return_if_fail(address_entry->input != NULL);
> 
>     if (!GTK_WIDGET_DRAWABLE(address_entry)) return;
> 
>     editable = GTK_EDITABLE(address_entry);
> 
>     input = address_entry->input;
>     show = g_string_new("");
>     cursor = start = end = 0;
>     found = FALSE;
>     for (list = g_list_first(input->list);
> 	 list != NULL;
> 	 list = g_list_next(list)) {
> 	/*
> 	 * Is it a normal string, or is it a match that requires ()
> 	 */
> 	addy = (emailData *)list->data;
> 	g_assert(addy != NULL);
> 	if (addy->match != NULL) {
> 	    out = g_strconcat("", addy->user, " (", addy->match, ")",
> NULL);
> 	} else {
> 	    out = g_strdup(addy->user);
> 	}
> 	/*
> 	 * Copy the string, adding a delimiter if need be.
> 	 */
> 	show = g_string_append(show, out);
> 	if (g_list_next(list) != NULL)
> 	    show = g_string_append(show, ", ");
> 
> 	/*
> 	 * Check for the cursor position.
> 	 */
> 	if (!found) {
> 	    if (list != input->active) {
> 		cursor += strlen(out);
> 	    	if (g_list_next(list) != NULL) cursor += 2;
> 	    } else {
> --->		found = TRUE;
> 		cursor += addy->cursor;
> 		if (addy->match) {
> 		    start = cursor - addy->cursor;
> 		    start += strlen(addy->user) + 1;
> 		    end = start + strlen(addy->match) + 2;
> 		}
> 	    }
> 	}
> 	g_free(out);
>     }
> ...
> 
> You see that in the line pointed by the arrow, you are in the else case,
> meaning that found is already TRUE, so you don't have to assign the TRUE
> value to it. Moreover found is always FALSE, because it is first
> initialized to FALSE.
> 
Damn I think I should have thought a bit more about this :-(
This is not a bug, it's correct, just forget that thing. Still the
remainder of the patch is correct (anyway it's really only one-liners).
Sorry
Bye
Manu