modifier default not working with @ suppression

[alabama]

The default modifier just access the tpl_vars array with the non existing key and tries to suppress the error outcome with '@' like:
$tmp = @$_smarty_tpl->tpl_vars['bShowInfo']->value

This won't work in PHP >= 5.4 (5.3 as well?)
There should be at least an "array_key_exist | isset" check to get the functionality right

[Miksser]

Is it possible to get rid of suppression in 'default'?
This behavior is very disturbing during debugging through xDebug =(

[Miksser]

Is it possible to get rid of suppression in 'default'?
This behavior is very disturbing during debugging through xDebug =(

[antman3351]

Had the same issue when debuging, I had a lots of notices in the console. Plus a possible performance loss from the notice being issued ( https://stackoverflow.com/questions/1868874/does-php-run-faster-without-warnings )

Maybe it's worth changing the original function?
Until then here's my workaround

/**
 * Overwrites smarty's default modifier
 * Original function used $tmp = @$_smarty_tpl->tpl_vars['param0']->value)===null||$tmp==='' ? $_smarty_tpl->tpl_vars[param1'']->value : $tmp)
 * Witch throws a notice that is silenced
 * @param array $params
 * @return mixed|string
 */
function smartyDefault( $params )
{
    $default = $params[1] ?? '';
    return "( isset({$params[0]}) && {$params[0]} !== '' ) ? {$params[0]} : {$default}";
}

 $smarty->registerPlugin('modifiercompiler', 'default', 'smartyDefault' );

[wisskid]

@Miksser @antman3351 I've added an empty check for variables for |default modifier in #650 . It does not appear to break anything. Can you confirm this fixes your problem?

[antman3351]

I just had a quick look at the change, so I might have missed something, but doesn't this change the behaviour of the default modifier?
before default was used for unset variable, null or an empty string. Using empty() default would also be applied to false, 0 , '0' and [ ].

[wisskid]

Well spotted, @antman3351 ! Not sure how that got past me. I thought I ran the new unit tests against the old code (seeing no change in behavior) but I must have somehow messed that up.
Added the 0, '0' and [] cases to the unit tests and fixed the implementation.
Is this better?

[wisskid]

The default modifier has already been changed to longer use the @ modifier in #629