Hi Dan and Internals Sorry Dan, I forgot to include @Internals in previous reply so let me resend this again.
Thank you for your reply. I see, but in that case it can be done with `$GLOBALS['abc']` right? So I don't see any benefits of allowing those forms, they're just another inconsistency that should not exist from the beginning. Yes, it does no harms but if nobody is really using it at all, is it good to remove this inconsistency? Or to make it really useful, why not just allow assignment like: global ${'abc'} = $someValue; Just my 2 cents. Regards On Sat, Sep 21, 2019 at 5:52 PM Nikita Popov <nikita....@gmail.com> wrote: > On Sat, Sep 21, 2019 at 11:56 AM Kosit Supanyo <webdevxp....@gmail.com> > wrote: > >> Hi Internals >> >> I'm working on my new proposals and I've found weirdness of global >> variable >> declaration in zend_language_parser.y. >> >> global_var: >> simple_variable >> { $$ = zend_ast_create(ZEND_AST_GLOBAL, >> zend_ast_create(ZEND_AST_VAR, $1)); } >> ; >> >> Above grammer allows something like this... >> >> global $$x; >> global $$$y; >> global $$$$z; >> global ${'abc'}; >> global $$$$$$$$$${random_int(0, PHP_INT_MAX)}; >> >> What's the propose of allowing this? And is there anyone out there knowing >> and using it? If not, should this be changed to allow only T_VARIABLE to >> make it consistent with `static` variable declaration? >> >> Regards >> > > Some grep results: > > sources/adodb/adodb-php/session/adodb-session.php > 676: global $$var; > 697: global $$var; > > sources/adodb/adodb-php/session/adodb-session2.php > 719: global $$var; > 740: global $$var; > > sources/adodb/adodb-php/session/old/adodb-cryptsession.php > 189: global $$var; > > sources/adodb/adodb-php/session/old/adodb-session.php > 300: global $$var; > > sources/adodb/adodb-php/session/old/adodb-session-clob.php > 269: global $$var; > > sources/wp-cli/wp-cli/php/WP_CLI/Runner.php > 1177: global ${$key}; > > > sources/apache/log4php/src/main/php/pattern/LoggerPatternConverterSuperglobal.php > 71: global ${$this->name}; > > We could deprecate this in favor of $GLOBALS, though as Dan said, the > motivation is not quite clear right now. > > Nikita >