rfc:null_coercion_consistency

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
rfc:null_coercion_consistency [2022/04/15 07:41] – craigfrancisrfc:null_coercion_consistency [2023/10/18 11:57] (current) – craigfrancis
Line 14: Line 14:
 ===== Introduction ===== ===== Introduction =====
  
-PHP 8.1 introduced [[https://wiki.php.net/rfc/deprecate_null_to_scalar_internal_arg|Deprecate passing null to non-nullable arguments of internal functions]]. While the consistency introduced by the RFC is welcome (user-defined vs internal functions), for those not using //strict_types=1//, the partial breaking of NULL coercion creates an upgrade problem.+PHP 8.1 introduced [[https://wiki.php.net/rfc/deprecate_null_to_scalar_internal_arg|Deprecate passing null to non-nullable arguments of internal functions]]. While the consistency is welcome (user-defined vs internal functions), for those **not** using strict Static Analysis or //strict_types=1//, the breaking of NULL coercion creates a difficult BC problem.
  
-This RFC does **not** change anything for scripts using //strict_types=1//, as type checks are useful in that context. For example, developers who view NULL as a missing/invalid value (not as a value in itself), consider passing NULL to a function like //htmlspecialchars()// as something that indicates a problem for them.+This RFC does **not** change anything for //strict_types=1// (or static analysis), as strict type checks can be useful. For example, some developers view NULL as a missing/invalid value (not as a value in itself), and passing NULL to a function like //htmlspecialchars()// might indicate a problem.
  
-Roughly **85%** scripts do not use //strict_types=1// (calculation below).+There was a [[https://externals.io/message/112327|short discussion]] about the original RFC, but with the exception of Craig Duncan, there was no discussion about the BC problems, or the inconsistency of NULL coercion compared to string/int/float/bool coercion, or other contexts like string concatenation, == comparisons, arithmetics, sprintf, print, echo, array keys, etc.
  
-Roughly **33%** of developers use static analysis (realistically it's less than this, details below). +The intention is to also keep [[https://github.com/Girgias/unify-typing-modes-rfc|Unify PHP's typing modes]] by George Peter Banyard in mind, with coercions like //substr($string, "offset")// and //htmlspecialchars(array())// as being clearly problematic; whereas the following is common:
- +
-There was a [[https://externals.io/message/112327|short discussion]] about the original RFC, but with the exception of Craig Duncan, there was no consideration for the problems this creates with existing code (or the inconsistency of NULL coercion compared to string/int/float/bool coercion). +
- +
-The general direction of [[https://github.com/Girgias/unify-typing-modes-rfc|Unify PHP's typing modes]] by Girgias is correct, because automatic coercions like //substr($string, "offset")// and //htmlspecialchars(array())// are clearly problematic; but the following is common, and has been fine:+
  
 <code php> <code php>
-$search = filter_input(INPUT_GET, 'q'); // Or similar (examples below)+$search = ($_GET['q'] ?? NULL); // Or similar (examples below)
  
-echo 'Results for ' . htmlspecialchars($search);+echo 'Results for ' . htmlspecialchars($search); // Fatal Error?!?
 </code> </code>
 +
 +Roughly **15%** of scripts use //strict_types=1// (calculation below).
 +
 +Roughly **33%** of developers use static analysis (realistically it's less than this, details below).
  
 ===== Problem ===== ===== Problem =====
Line 45: Line 45:
   - [[https://www.php.net/manual/en/language.types.boolean.php|To Boolean]]: "When converting to bool, the following values are considered false [...] the special type NULL"   - [[https://www.php.net/manual/en/language.types.boolean.php|To Boolean]]: "When converting to bool, the following values are considered false [...] the special type NULL"
  
-For example:+String/Int/Float/Bool can be coerced; Arrays, Resources, and Objects (without toString) cannot be coerced (for fairly obvious reasons). 
 + 
 +Or, NULL "can be coerced to the type requested by the hint without data loss and without creation of likely unintended data" ([[https://wiki.php.net/rfc/coercive_sth|ref]]). 
 + 
 +PHP 7.0 introduced the ability for user-defined functions to specify parameter types via the [[https://wiki.php.net/rfc/scalar_type_hints_v5#behaviour_of_weak_type_checks|Scalar Type Declarations RFC]], where the implementation triggered Type Errors for those using //strict_types=1//, and otherwise used coercion for string/int/float/bool, but not NULL, PHP 8.1 updated internal function parameters to work in the same way. 
 + 
 +==== Current State ====
  
 <code php> <code php>
Line 52: Line 58:
 print(1.2); print(1.2);
 print(false); print(false);
-print(NULL); // Fine, coerced to empty string.+print(NULL); // Fine, still coerced to empty string. 
 + 
 +echo NULL; // Fine, still coerced to empty string. 
 + 
 +var_dump(3 + '5' + NULL); // Fine, int(8) 
 +var_dump(NULL / 6); // Fine, int(0)
  
 $o = []; $o = [];
  
 $o[] = ('' == ''); $o[] = ('' == '');
-$o[] = ('' == NULL); // Fine, coerced to empty string.+$o[] = ('' == NULL); // Fine, still coerced to empty string.
  
 $o[] = 'ConCat ' . 'A'; $o[] = 'ConCat ' . 'A';
Line 63: Line 74:
 $o[] = 'ConCat ' . 1.2; $o[] = 'ConCat ' . 1.2;
 $o[] = 'ConCat ' . false; $o[] = 'ConCat ' . false;
-$o[] = 'ConCat ' . NULL; // Fine, coerced to empty string.+$o[] = 'ConCat ' . NULL; // Fine, still coerced to empty string.
  
 $o[] = sprintf('%s', 'A'); $o[] = sprintf('%s', 'A');
Line 69: Line 80:
 $o[] = sprintf('%s', 1.2); $o[] = sprintf('%s', 1.2);
 $o[] = sprintf('%s', false); $o[] = sprintf('%s', false);
-$o[] = sprintf('%s', NULL); // Fine, coerced to empty string.+$o[] = sprintf('%s', NULL); // Fine, still coerced to empty string.
  
 $o[] = htmlspecialchars('A'); $o[] = htmlspecialchars('A');
Line 78: Line 89:
 </code> </code>
  
-With user-defined functions, while this does not cause a backwards compatibility issue (details below), it still highlights the coercion inconsistency, and that some type checking is happening in an environment that does not expect type checking:+With user-defined functions, this inconsistency is also noted:
  
 <code php> <code php>
Line 92: Line 103:
  
 user_function(3.3, 3.3, 3.3, 3.3); user_function(3.3, 3.3, 3.3, 3.3);
-  // string(3) "3.3" / int(3) / float(3.3) / bool(true)+  // string(3) "3.3" / int(3), lost precision / float(3.3) / bool(true)
  
 user_function(false, false, false, false); user_function(false, false, false, false);
Line 98: Line 109:
  
 user_function(NULL, NULL, NULL, NULL); user_function(NULL, NULL, NULL, NULL);
-  // Uncaught TypeError x4?+  // Fatal error, Uncaught TypeError x4!
 </code> </code>
  
-Arrays, Resources, and Objects (without //__toString()//) cannot be coerced (for fairly obvious reasons).+==== Scalar Types ====
  
-String/Int/Float/Bool can be coerced.+The [[https://wiki.php.net/rfc/scalar_type_hints_v5|Scalar Type Declarations]] RFC says "it should be possible for existing userland libraries to add scalar type declarations without breaking compatibility", but this is not the case, because of NULL. This has made adoption of type declarations harder, as it does not work like the following:
  
-NULL can usually be coerced, but...+<code php> 
 +function my_function($s, $i, $f, $b) { 
 +  $s = strval($s); 
 +  $i = intval($i); 
 +  $f = floatval($f); 
 +  $b = boolval($b); 
 +  var_dump($s, $i, $f, $b); 
 +}
  
-  - PHP 7.0 introduced the ability for user-defined functions to specify parameter types via the [[https://wiki.php.net/rfc/scalar_type_hints_v5#behaviour_of_weak_type_checks|Scalar Type Declarations RFC]], where the focus was on //strict_types=1//. But the implementation also caused Type Errors when not using //strict_types=1//, which seems more of an over-sight, with coercion working for string/int/float/bool but not null (despite the documentation), and introducing a type check (for an environment that does not expect this). +function my_function(string $s, int $i, float $f, bool $b) { 
-  - PHP 8.1 continued this inconsistency with internal functions.+  var_dump($s, $i, $f, $b); 
 +} 
 + 
 +my_function(NULL, NULL, NULL, NULL); 
 +</code> 
 + 
 +[[https://news-web.php.net/php.internals/117523|George Peter Banyard]] notes that "Userland scalar types [...] did not include coercion from NULL for //very// good reasons". 
 + 
 +The [[https://wiki.php.net/rfc/scalar_type_hints_v5|Scalar Type Declarations]] RFC says "The only exception to this is the handling of NULL: in order to be consistent with our existing type declarations for classes, callables and arrays, NULL is not accepted by default, unless it is a parameter and is explicitly given a default value of NULL", which goes against the documentation (as noted above) where the coercion from NULL is well defined (i.e. NULL is more like a string/int/float/bool, rather than an object/callable/array).
  
 ==== Examples ==== ==== Examples ====
Line 128: Line 154:
 $value = array_pop($empty_array); $value = array_pop($empty_array);
 $value = mysqli_fetch_row($result); $value = mysqli_fetch_row($result);
-$value = json_decode($json); // Invalid JSON, or nesting limit. 
 </code> </code>
  
-Examples, often working with user input, where NULL has previously been fine:+Examples functions, often working with user input, where NULL has always been accepted/coerced:
  
 <code php> <code php>
 +$rounded_value = round($value);
 +
 $search_trimmed = trim($search); $search_trimmed = trim($search);
  
Line 164: Line 191:
 mail('nobody@example.com', 'subject', 'message', NULL, '-fwebmaster@example.com'); mail('nobody@example.com', 'subject', 'message', NULL, '-fwebmaster@example.com');
 </code> </code>
 +
 +HTML Templating engines like [[https://github.com/laravel/framework/blob/ab1506091b9f166b312b3990d07b2e21d971f2e6/src/Illuminate/Support/helpers.php#L119|Laravel Blade]] suppress this deprecation with null-coalescing ([[https://github.com/laravel/framework/pull/36262/files#diff-15b0a3e2eb2d683222d19dfacc04c616a3db4e3d3b3517e96e196ccbf838f59eR118|patch]]); or [[https://github.com/twigphp/Twig/blob/b4d6723715da57667cca851051eba3786714290d/src/Extension/EscaperExtension.php#L195|Symphony Twig]] which preserves NULL, but it's often passed to //echo// (which accepts it, despite the [[https://www.php.net/echo|echo documentation]] saying it accepts non-nullable strings).
 +
 +I'd argue strict type checking (that prevents all forms of coercion) should be done via Static Analysis or via //strict_types=1// opt-in, like how a string (e.g. '15') being provided to integer parameter could identify a problem in some development environments.
  
 There are approximately [[https://github.com/craigfrancis/php-allow-null-rfc/blob/main/functions-change.md|335 parameters affected by this deprecation]]. There are approximately [[https://github.com/craigfrancis/php-allow-null-rfc/blob/main/functions-change.md|335 parameters affected by this deprecation]].
Line 172: Line 203:
  
 The only realistic way for developers to find when NULL is passed to these internal functions is to use the deprecation notices (not ideal). The only realistic way for developers to find when NULL is passed to these internal functions is to use the deprecation notices (not ideal).
 +
 +And this does not stop new issues being introduced, as developers may not consider when variables might be NULL (e.g. a browser extension or network issues affecting GET/POST variables).
  
 It is possible to use very strict Static Analysis, to follow every variable from source to sink (to check if a variable could be NULL), but most developers are not in a position to do this (i.e. not using static analysis, or not at a high enough level, or they are using a baseline to ignore). It is possible to use very strict Static Analysis, to follow every variable from source to sink (to check if a variable could be NULL), but most developers are not in a position to do this (i.e. not using static analysis, or not at a high enough level, or they are using a baseline to ignore).
  
-In the last JetBrains developer survey, where 67% regularly used Laravel, **only 33% used Static Analysis** ([[https://www.jetbrains.com/lp/devecosystem-2021/php/#PHP_do-you-use-static-analysis|source]]); where it's fair to say many still would not identify these possible NULL values.+In the last JetBrains developer survey (with 67% regularly using Laravel), **only 33% used Static Analysis** ([[https://www.jetbrains.com/lp/devecosystem-2021/php/#PHP_do-you-use-static-analysis|source]]); where it's fair to say many of these developers would //still// not identify these possible NULL values (too low level, and/or using a baseline).
  
 As an example, take this simple script: As an example, take this simple script:
Line 187: Line 220:
 </code> </code>
  
-It's considered fine today by the relevant tools: +This is considered fine by these tools:
- +
-<code cli> +
-composer require --dev vimeo/psalm +
-./vendor/bin/psalm --init ./src/ 4 +
-./vendor/bin/psalm +
-No errors found! +
-</code> +
-Note: Psalm can detect this problem at [[https://psalm.dev/docs/running_psalm/error_levels/|levels 1, 2, and 3]] (don't use a baseline). +
- +
-<code cli> +
-composer require --dev phpstan/phpstan +
-./vendor/bin/phpstan analyse -l 9 ./src/ +
-[OK] No errors +
-</code> +
- +
-<code cli> +
-composer require --dev phpstan/phpstan-strict-rules +
-composer require --dev phpstan/extension-installer +
-./vendor/bin/phpstan analyse -l 9 ./src/ +
-[OK] No errors +
-</code> +
-Note: There are [[https://phpstan.org/config-reference#stricter-analysis|Stricter Analysis]] options for PHPStan, but they don't seem to help with this problem. +
- +
-<code cli> +
-composer require --dev rector/rector +
-./vendor/bin/rector init +
-./vendor/bin/rector process ./src/ +
-[OK] Rector is done! +
-</code>+
  
 <code cli> <code cli>
Line 259: Line 263:
  
 Note: Juliette (@jrfnl) has confirmed that getting PHPCompatibility to solve this problem will be "pretty darn hard to do" because it's "not reliably sniffable" ([[https://twitter.com/jrf_nl/status/1497937320766496772|source]]). Note: Juliette (@jrfnl) has confirmed that getting PHPCompatibility to solve this problem will be "pretty darn hard to do" because it's "not reliably sniffable" ([[https://twitter.com/jrf_nl/status/1497937320766496772|source]]).
 +
 +<code cli>
 +composer require --dev phpstan/phpstan
 +./vendor/bin/phpstan analyse -l 9 ./src/
 +[OK] No errors
 +</code>
 +
 +<code cli>
 +composer require --dev phpstan/phpstan-strict-rules
 +composer require --dev phpstan/extension-installer
 +./vendor/bin/phpstan analyse -l 9 ./src/
 +[OK] No errors
 +</code>
 +Note: There are [[https://phpstan.org/config-reference#stricter-analysis|Stricter Analysis]] options for PHPStan, but they don't seem to help with this problem.
 +
 +<code cli>
 +composer require --dev vimeo/psalm
 +./vendor/bin/psalm --init ./src/ 4
 +./vendor/bin/psalm
 +No errors found!
 +</code>
 +Note: Psalm can detect this at [[https://psalm.dev/docs/running_psalm/error_levels/|levels 1, 2, and 3]] (don't use a baseline).
 +
 +==== One Solution ====
 +
 +Since [[https://github.com/rectorphp/rector-src/pull/2543|21st June 2022]], Rector can modify 362 function arguments via //NullToStrictStringFuncCallArgRector//:
 +
 +<code cli>
 +mkdir -p rector/src;
 +
 +cd rector/;
 +
 +composer require --dev rector/rector;
 +
 +echo '<?= htmlspecialchars($var) ?>' > src/index.php;
 +
 +echo '<?php
 +
 +use Rector\Php81\Rector\FuncCall\NullToStrictStringFuncCallArgRector;
 +use Rector\Config\RectorConfig;
 +
 +return static function (RectorConfig $rectorConfig): void {
 +    $rectorConfig->paths([__DIR__ . "/src"]);
 +    $rectorConfig->rule(NullToStrictStringFuncCallArgRector::class);
 +};
 +' > rector.php;
 +
 +./vendor/bin/rector process;
 +</code>
 +
 +This will litter the code with the use of //(string)// type casting, e.g.
 +
 +<code diff>
 +-<?= htmlspecialchars($var) ?>
 ++<?= htmlspecialchars((string) $var) ?>
 +</code>
 +
 +For a typical project (which won't be using //strict_types=1//), expect thousands of changes to be made; and note how this does not improve code quality.
  
 ==== Temporary Solutions ==== ==== Temporary Solutions ====
Line 281: Line 343:
 ==== Updating ==== ==== Updating ====
  
-While making each change is fairly easy - they are difficult to find, there are many of them (time consuming), and the updates used are often pointless, e.g.+While making each change is fairly easy - they are still difficult to find, there are many of them, and the updates used are often pointless, e.g.
  
   * //urlencode(strval($name));//   * //urlencode(strval($name));//
Line 287: Line 349:
   * //urlencode($name ?? "");//   * //urlencode($name ?? "");//
  
-One example diff didn't exactly make the code easier to read:+And often do not make the code easier to read:
  
 <code diff> <code diff>
Line 294: Line 356:
 </code> </code>
  
-As noted above - PHPCompatibility, CodeSniffer, Rector, etc are unable to find or update these cases.+As noted above - Rector can add //(string)// type casting automatically, but this does not improve code quality.
  
 ===== Proposal ===== ===== Proposal =====
  
-No change for those using //strict_types=1//.+Revert the NULL deprecation for parameters (when **not** using //strict_types=1//), so NULL coercion continues to work (as it does in other contexts) to avoid the upgrade problems (i.e. Fatal Errors in PHP 9.0).
  
-Must keep the spirit of the original RFC, and keep user-defined and internal functions consistent.+And, in the spirit of the original RFC to keep user-defined and internal functions consistent, also change user-defined functions so NULL is coerced for non-nullable parameters (when **not** using //strict_types=1//).
  
-Revert the deprecation of NULL coercion when **not** using //strict_types=1//, to work as documented, be consistent with scalar types, and (most importantly) avoid the upgrade problems (i.e. Fatal Errors for PHP 9.0). +This means the type "//?int//" will allow NULL or an integer to be provided to the function; whereas the non-nullable type "//int//" would coerce NULL to 0; in exactly the same way the string "0" or the boolean false would be.
- +
-For consistency, change user-defined functions to allow NULL to be coerced when **not** using //strict_types=1//.+
  
 ===== Backward Incompatible Changes ===== ===== Backward Incompatible Changes =====
  
-None+While the intention of this RFC is to avoid a BC break; for user defined functions to be updated to also coerce NULL (instead of throwing a Type Error), while unlikely, it is possible some non //strict_types// code may rely the TypeError being thrown, for example: 
 + 
 +<code php> 
 +function my_function(string $my_string) { 
 +  var_dump($my_string); 
 +} 
 + 
 +try { 
 +  my_function('A');   // string(1) "A" 
 +  my_function(1);     // string(1) "1" 
 +  my_function(1.2);   // string(3) "1.2" 
 +  my_function(true);  // string(1) "1" 
 +  my_function(false); // string(0) "" 
 +  my_function(NULL);  // Throw Type Error 
 +} catch (TypeError $e) { 
 +  // Do something important? 
 +} 
 +</code>
  
 ===== Proposed PHP Version(s) ===== ===== Proposed PHP Version(s) =====
Line 329: Line 406:
  
 ===== Open Issues ===== ===== Open Issues =====
- 
-"terrible idea" - I'm still waiting to hear details. 
  
 "it's a bit late" - We only have a deprecation at the moment (which can and is being ignored), it will be "too late" when PHP 9.0 uses Fatal Errors. "it's a bit late" - We only have a deprecation at the moment (which can and is being ignored), it will be "too late" when PHP 9.0 uses Fatal Errors.
  
-"Userland scalar types [...] did not include coercion from NULL for //very// good reasons." - The only reason mentioned in [[https://wiki.php.net/rfc/scalar_type_hints_v5|Scalar Type Declarations]] is "to be consistent with our existing type declarations" (no further details given). Talking to developers, the only reason mentioned is noted above, where NULL can be viewed as a missing/invalid value, and passing NULL to a function like //htmlspecialchars()// could indicate a problem (while that might be useful in the context of //strict_types=1//, it hasn't been the case for everyone else).+The function //mt_rand()// can be called with no arguments, or with min and max integer arguments. A developer may call //mt_rand(NULL, NULL)// and expect it to work the same as no arguments (returning a random number between 0 and //mt_getrandmax()//), but the NULL's would be coerced to 0, so it would always return 0. That said, I cannot find any public examples of this happening ([[https://grep.app/search?q=mt_rand%28NULL&filter%5Blang%5D%5B0%5D=PHP|1]], [[https://grep.app/search?q=mt_rand%5Cs%2A%5C%28%5Cs%2ANULL&regexp=true&filter[lang][0]=PHP|2]], [[https://www.google.com/search?q=%22mt_rand+NULL+NULL%22|3]]).
  
 ===== Future Scope ===== ===== Future Scope =====
Line 344: Line 419:
   - //$method// in [[https://php.net/method_exists|method_exists()]]   - //$method// in [[https://php.net/method_exists|method_exists()]]
   - //$json// in [[https://php.net/json_decode|json_decode()]]   - //$json// in [[https://php.net/json_decode|json_decode()]]
 +
 +It might be appropriate for coercion and explicit casting/converting to work in the same way, even if they were to become stricter in the values they accept; e.g. //intval("")// and //((int) "")// currently return int(0), whereas //(5 + "")// results in a TypeError.
  
 ===== Voting ===== ===== Voting =====
Line 361: Line 438:
 ===== Notes ===== ===== Notes =====
  
-The **85%** of scripts that do not use //strict_types=1// was calculated using [[https://grep.app/|grep.app]], to "search across a half million git repos", were each result is a script (not a count of matches,  [[https://grep.app/search?q=defuse/php-encryption&filter[lang][0]=PHP|example]]). We can see [[https://grep.app/search?q=strict_types&filter[lang][0]=PHP|269,701]] scripts using //strict_types=1//, out of [[https://grep.app/search?q=php&filter[lang][0]=PHP|1,830,411]]. And keep in mind that [[https://grep.app/search?q=class%20wpdb%20%7B&filter[lang][0]=PHP|WordPress only really appears once]], it is [[https://make.wordpress.org/core/2022/01/10/wordpress-5-9-and-php-8-0-8-1/#php-8-1-deprecation-passing-null-to-non-nullable-php-native-functions-parameters|affected by this deprecation]], and is installed/used by many.+The **15%** of scripts that use //strict_types=1// was calculated using [[https://grep.app/|grep.app]], to "search across a half million git repos", were each result is a script (not a count of matches,  [[https://grep.app/search?q=defuse/php-encryption&filter[lang][0]=PHP|example]]). We can see [[https://grep.app/search?q=strict_types&filter[lang][0]=PHP|272,871]] scripts using //strict_types=1//, out of [[https://grep.app/search?q=php&filter[lang][0]=PHP|1,842,666]]. And keep in mind that [[https://grep.app/search?q=class%20wpdb%20%7B&filter[lang][0]=PHP|WordPress only really appears once]], it is [[https://make.wordpress.org/core/2022/01/10/wordpress-5-9-and-php-8-0-8-1/#php-8-1-deprecation-passing-null-to-non-nullable-php-native-functions-parameters|affected by this deprecation]], and is installed/used by many.
  
 In the [[https://wiki.php.net/rfc/scalar_type_hints_v5#behaviour_of_weak_type_checks|Scalar Type Declarations]] RFC for PHP 7.0, scalar types were defined as "int, float, string and bool" - but, despite NULL also being a simple value (i.e. not an array/object/resource), it was not included in this definition. For backwards compatibility reasons this definition is unlikely to change. In the [[https://wiki.php.net/rfc/scalar_type_hints_v5#behaviour_of_weak_type_checks|Scalar Type Declarations]] RFC for PHP 7.0, scalar types were defined as "int, float, string and bool" - but, despite NULL also being a simple value (i.e. not an array/object/resource), it was not included in this definition. For backwards compatibility reasons this definition is unlikely to change.
rfc/null_coercion_consistency.1650008466.txt.gz · Last modified: 2022/04/15 07:41 by craigfrancis

Page Tools