rfc:readline_interactive_shell_result_function

This is an old revision of the document!


PHP RFC: Dump results of expressions in `php -a`

Introduction

Many REPLs (Read-Eval-Print Loops) for other programming languages that I'm familiar with print a (possibly truncated) representation of the result of expressions, but PHP doesn't. It would be useful to allow users to extend the functionality of the default interactive php shell (php -a), possibly with auto_prepend_file or through use of command wrappers/aliases, or after loading the shell. Prior to this RFC, there was no way to extend the interactive php shell in this way. (I've seen https://github.com/bobthecow/psysh mentioned as an alternative for php -a while investigating this, but that's a shell written from scratch, and doesn't have some functionality from php -a such as tolerance of fatal errors)

Proposal

By default, in php -a, dump the results of non-null expressions using var_dump/var_export(). Additionally, add a new function readline_interactive_shell_result_function to the readline PHP module. This function only affects interactive shells - it can be used to set or clear a closure when extension_loaded('readline') === true, but that closure would only be called in interactive shells (i.e. php -a).

This will dump the results of expressions every time a statement containing a single expression such as 2+2; or $x = call_function(); is evaluated (but not non-expressions such as class X{}, statement blocks such as { $x = 1; }, or combinations of expressions such as $x = 1; $y = $x*2;.

An example of the behavior of the default expression dumper is below:

$ php -a
Interactive shell
 
php > 1+1;
=> 2
php > 0.5 * 2;
=> 1.0
php > namespace\MyClass::class;
=> 'MyClass'
php > fn()=>true;
=> object(Closure)#1 (0) {
}
php > $x = ["foo", "bar"];
=> array(2) {
  [0]=>
  string(3) "foo"
  [1]=>
  string(3) "bar"
}
php > asort($x);
=> true
php > $x;
=> array(2) {
  [1]=>
  string(3) "bar"
  [0]=>
  string(3) "foo"
}
php > json_encode($x);
=> '{"1":"bar","0":"foo"}'
php > unset($x);
php > function do_something() { echo "in do_something()\n"; }
php > do_something();
in do_something()
php > json_decode('{"key": "value"}');
=> object(stdClass)#1 (1) {
  ["key"]=>
  string(5) "value"
}
php > throw new RuntimeException("test");
 
Warning: Uncaught RuntimeException: test in php shell code:1
Stack trace:
#0 {main}
  thrown in php shell code on line 1
php > printf("newline is automatically appended by shell");
newline is automatically appended by shell
=> 42
php > printf("newline not automatically appended by shell\n");
newline not automatically appended by shell
=> 44
php > { print("test\n"); } // statement blocks can be used to avoid dumping expression results
test
php >

Using the function readline_interactive_shell_result_function(), users or applications can override the default expression dumper, or remove it entirely.

$ php -a
Interactive shell
 
php > readline_interactive_shell_result_function(
php (     function (string $code, $result) {
php (         echo "Saw " . trim($code) . "\n";
php (         echo json_encode($result);
php (     });
Saw readline_interactive_shell_result_function(
     function (string $code, $result) {
         echo "Saw " . trim($code) . "\n";
         echo json_encode($result);
     });
true
php > 2+2;
Saw 2+2;
4
php > readline_interactive_shell_result_function(null);  // useful if output would be extremely long
php > 2+2;
php > 

Real implementations may be much more complex, and make use of parsers such as https://github.com/nikic/PHP-Parser or https://github.com/nikic/php-ast to check if the expression in question is an assignment, call to print(), etc.

This has the following signature:

/**
 * When $callback is a callback, replaces the old callback used to dump expression results.
 * When $callback is null, removes the callback used to dump expression results.
 *
 * Currently, this always returns true, but future changes to the implementation
 * may make it return false.
 */
function readline_interactive_shell_result_function(?callable $callback): bool;

The default implementation added as part of this RFC is effectively identical to the below implementation, but written in C. It can be replaced with a userland implementation or disabled at any time, even in auto_prepend_file. (Because the default implementation is written in C, it will work even if the ini setting disable_functions includes var_dump and var_export.)

readline_interactive_shell_result_function(
    function(string $code, $result) {
        if (!isset($result)) {
            return;
        }
        if (is_scalar($result)) {
            echo "=> " . var_export($result, true) . "\n";
        } else {
            echo "=> "; var_dump($result);
        }});

A new system ini boolean setting cli.enable_interactive_shell_result_function is added as part of this RFC, defaulting to enabled. It can be disabled to prevent expression results from being dumped at all (by the internal implementation or by overrides), e.g. if the closures have bugs, are excessivly verbose, or have unpredictable performance.

Backward Incompatible Changes

Only interactive sessions (php -a) are affected, by a difference in the output sent to stdout. The dumping of expression results can be disabled entirely with the ini setting cli.enable_interactive_shell_result_function = Off, or temporarily by calling readline_interactive_shell_result_function(null)

In interactive sessions, this will start calling __debugInfo() if it exists due to calling var_dump() on objects. Implementations of __debugInfo() may throw or have other side effects after the expression is evaluated.

Proposed PHP Version(s)

8.1

RFC Impact

To SAPIs

This functionality is available in interactive CLI sessions. Other SAPIs are unaffected.

php.ini Defaults

If there are any php.ini settings then list:

  • hardcoded default values: cli.enable_interactive_shell_result_function = On
  • php.ini-development values: cli.enable_interactive_shell_result_function = On (or omitted, cli.pager and cli.prompt currently exist but not documented in those files)
  • php.ini-production values: cli.enable_interactive_shell_result_function = On

Future Scope

From https://externals.io/message/111073#111073

Miscellaneous thoughts on implementation details:

by being less reliant on heuristics (e.g. checking if the token $var was a variable or a property, making it easier to collect local variables, etc).

Is packaging a parser practical for a phpi binary (e.g. for package managers, maintainers of php, other reasons)?

  • A parser may fail for code using new token types until the parser gets updated to handle the new token types. This stops being a concern after feature freezes.
    Looping over @token_get_all() and bailing out on an unknown token type may help with that.
  • How would crash/bug fixes of phpi or the parser be handled in patch releases of php if this was released with php?
  • Automatically rewriting the code to namespace the parser and its dependencies with \PHP\Internal\BundledPhpParser would let phpi be used with projects that depend on a different php-parser version.
    (clarifications may be necessary to indicate to end users that the bundled parser copy won't get updates or support outside of php minor releases, should not be used by libraries/applications and that it won't support newer php syntax, and possibly other things)

Supporting rendering the last expression in a list of statements

This was left out to simplify the implementation, but is probably doable by rewriting the C AST before evaluating the snippet for safe node kinds.

  • i.e. change foo(); bar(); to foo(); return (bar());

Rendering a result limit

https://externals.io/message/112934#113039 was brought up after the start of the vote. While some REPLs don't limit results, others do, and it's useful to have for extremely large objects. The setting cli.pager can be used to limit results, but is disabled by default.

My main concern in this iteration of the RFC is: what happens with big/deeply nested objects? They tend to spew tons of lines if var_dump()'ed. Do we have reasonable depth/output limitations in default dumping mode?

I'm often enough using php -a to do some quick ad-hoc processing (example, read a big json file, and then access a value; instantiating a mediawiki bot framework and calling replace on it; ...).

It's really cool to have any interactive feedback at all, but please, at least by default, limit the output. (An example is the JS REPL in browser console - it shows you a minimal preview of the object, and then you can expand with your mouse. Obviously with a pure cli application, this needs different - intuitive - navigation.)

As it currently stands, this makes php -a unusable in any but the simplest cases, without just disabling the whole feature.

I like the whole feature, but the missing output limitation (I have yet enough nightmares from var_dump()'ing the wrong object filling my shell with tons of irrelevant information… I don't need that potentially happening on every single evaluated expression)

Thus I'm voting no, for now.

Vote

Voting starts on 2021-01-19 and ended 2021-02-02

Yes/No, requiring 2/3 majority

Dump results of expressions in `php -a` as described in this RFC
Real name Yes No
as (as)  
asgrim (asgrim)  
ashnazg (ashnazg)  
bwoebi (bwoebi)  
cmb (cmb)  
galvao (galvao)  
jasny (jasny)  
kalle (kalle)  
kinncj (kinncj)  
nikic (nikic)  
ramsey (ramsey)  
reywob (reywob)  
sergey (sergey)  
svpernova09 (svpernova09)  
tandre (tandre)  
theodorejb (theodorejb)  
villfa (villfa)  
Count: 11 6

References

Changelog

0.2: Dump non-null expression results by default with var_dump()/var_export() 0.3: Document the default implementation used in the implementation

rfc/readline_interactive_shell_result_function.1612312712.txt.gz · Last modified: 2021/02/03 00:38 by tandre