# [RFC][Discussion] Make compact function reports undefined passed variables - Thread: https://externals.io/message/102012 - Original archive: https://news-web.php.net/php.internals/102012 - Started by Gabriel Caruso on 2018-04-02T09:17:43+00:00 - Messages: 7 - Reply by email to internals@lists.php.net - More for agents: search with https://externals.io/search.md?q={query} · JSON API at https://externals.io/api · documentation at https://externals.io/llms.txt --- ## #102012 — Gabriel Caruso — 2018-04-02T09:17:43+00:00 URL: https://externals.io/message/102012#102012 Hello dear internals, how are you? I'd like to propose a new RFC to PHP's core, but as this one contains a BC Break, let's discussed it before making anything official. A couple of days ago, while discussing some [Coding Standards rules for Doctrine, forbidden the `compact` function]( https://github.com/doctrine/coding-standard/pull/49), an argument caught my attention: > The `compact` function var does not report undefined variables. Looking in [the `compact` documentation](https://secure.php.net/compact), this is even emphasizes: > Any strings that are not set will simply be skipped. I couldn't figure out why this is done this way, but, here's what I'd like to propose: make the `compact` function starts to report undefined passed variables for it. With [only 2 lines of code]( https://github.com/php/php-src/compare/master...carusogabriel:warning-unknow-compact-variable), this is possible, but, of course, this is a BC Break. Let me know your opinion on that, and perhaps, make it happen! Regards, -- Gabriel Caruso --- ## #102013 — Pieter Hordijk — 2018-04-02T11:21:47+00:00 URL: https://externals.io/message/102012#102013 · In reply to: #102012 ----- Original Message ----- > From: "Gabriel Caruso" > To: "internals" > Sent: Monday, April 2, 2018 11:17:43 AM > Subject: [PHP-DEV] [RFC][Discussion] Make compact function reports undefined passed variables > Hello dear internals, how are you? > > I'd like to propose a new RFC to PHP's core, but as this one contains a BC > Break, let's discussed it before making anything official. > > A couple of days ago, while discussing some [Coding Standards rules for > Doctrine, forbidden the `compact` function]( > https://github.com/doctrine/coding-standard/pull/49), an argument caught my > attention: > > > […] > > Looking in [the `compact` documentation](https://secure.php.net/compact), > this is even emphasizes: > > > […] > > I couldn't figure out why this is done this way, but, here's what I'd like > to propose: make the `compact` function starts to report undefined passed > variables for it. > > With [only 2 lines of code]( > https://github.com/php/php-src/compare/master...carusogabriel:warning-unknow-compact-variable), > this is possible, but, of course, this is a BC Break. > > Let me know your opinion on that, and perhaps, make it happen! > > Regards, > -- > Gabriel Caruso Hey Gabriel, What I am missing here is the reason why we should break well defined behavior of existing functions. How is the BC break for all existing code justified? Pieter --- ## #102014 — Gabriel Caruso — 2018-04-02T11:45:47+00:00 URL: https://externals.io/message/102012#102014 · In reply to: #102013 > > Hey Gabriel, > > What I am missing here is the reason why we should break well defined > behavior of existing functions. > > How is the BC break for all existing code justified? > > Pieter > Hi Pieter. > What I am missing here is the reason why we should break well defined behavior of existing functions. This is something I'd like to discuss about `compact`. I'd like to know why we skip undefined variables instead of report them. `compact` is use in a large scale by the MVC community, by passing variables in the controller and using them in the view. > How is the BC break for all existing code justified? I'd like to start report this undefined variables. Would be one more way to prevent code-smell. IMHO this belongs to the core, not to the userland. Thanks, -- Gabriel Caruso --- ## #102015 — Pieter Hordijk — 2018-04-02T12:08:50+00:00 URL: https://externals.io/message/102012#102015 · In reply to: #102014 > Hi Pieter. > > […] > Hi Pieter. > > […] > This is something I'd like to discuss about `compact`. I'd like to know why we > skip undefined variables instead of report them. `compact` is use in a large > scale by the MVC community, by passing variables in the controller and using > them in the view. > > […] > I'd like to start report this undefined variables. Would be one more way to > prevent code-smell. IMHO this belongs to the core, not to the userland. > Thanks, > -- > Gabriel Caruso Regardless of the why it is like that the fact of the matter is that it is and imho you would need a really good reason to break existing code. I still don't don't see your reason for it. Nor do I see how it is a code smell atm. --- ## #102016 — Ivan Enderlin — 2018-04-04T07:32:43+00:00 URL: https://externals.io/message/102012#102016 · In reply to: #102012 Hello, Yes, a thousand yes. Silently ignoring undefined variables is a source of multiple bugs. Emiting a warning is a small BC break comparing to the benefits. Regards. > Hello dear internals, how are you? > > I'd like to propose a new RFC to PHP's core, but as this one contains a BC > Break, let's discussed it before making anything official. > > A couple of days ago, while discussing some [Coding Standards rules for > Doctrine, forbidden the `compact` function]( > https://github.com/doctrine/coding-standard/pull/49), an argument caught my > attention: > > > […] > Looking in [the `compact` documentation](https://secure.php.net/compact), > this is even emphasizes: > > > […] > I couldn't figure out why this is done this way, but, here's what I'd like > to propose: make the `compact` function starts to report undefined passed > variables for it. > > With [only 2 lines of code]( > https://github.com/php/php-src/compare/master...carusogabriel:warning-unknow-compact-variable), > this is possible, but, of course, this is a BC Break. > > Let me know your opinion on that, and perhaps, make it happen! > > Regards, --- ## #102026 — Andrea Faulds — 2018-04-09T21:54:50+00:00 URL: https://externals.io/message/102012#102026 · In reply to: #102016 And the determined can prefix with an @ if they want similar to old behaviour :) Ivan Enderlin wrote: > Hello, > > Yes, a thousand yes. Silently ignoring undefined variables is a source > of multiple bugs. Emiting a warning is a small BC break comparing to the > benefits. > > Regards. > -- Andrea Faulds https://ajf.me/ --- ## #102027 — bjorn.x.larsson@telia.com — 2018-04-11T13:49:56+00:00 URL: https://externals.io/message/102012#102027 · In reply to: #102026 And by that one could argue that the BC break is negligible :) Anyway, I think this is a clear improvement! Regards -- //Björn Larsson Den 2018-04-09 kl. 23:54, skrev Andrea Faulds: > And the determined can prefix with an @ if they want similar to old > behaviour :) > > Ivan Enderlin wrote: > > […]