Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:108042
Content-Type: text/plain;
	charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.11\))
In-Reply-To: <5455D4D6-9B07-4957-BA98-489EE12FEDD8@newclarity.net>
Date: Wed, 8 Jan 2020 17:50:19 -0500
Cc: Rowan Tommins <rowan.collins@gmail.com>,
 PHP internals <internals@lists.php.net>
Content-Transfer-Encoding: quoted-printable
Message-ID: <B9DF9814-A09F-424C-AF57-001FC13F8394@newclarity.net>
References: <5e0d723f.1c69fb81.e2ae8.24e2SMTPIN_ADDED_MISSING@mx.google.com>
 <74F2DBFC-E63C-428C-A37F-2D0CEE15AD0F@newclarity.net>
 <DM5PR07MB30676FC66DA036895D5E3554F9200@DM5PR07MB3067.namprd07.prod.outlook.com>
 <CD014984-715D-49C5-9A90-66C0A11F9818@newclarity.net>
 <c40a96fe-8929-4b44-98bd-3370a12ec516@www.fastmail.com>
 <D56361F6-8B11-4363-908D-962F683936BF@newclarity.net>
 <b32edf0c-451d-2281-8036-0b631a3c7a6f@gmail.com>
 <CD3AFCD7-EF2F-46C7-90D3-3A83A1D8836C@newclarity.net>
 <53556dfb-44ce-f902-204c-9a7da9484a61@gmail.com>
 <65567C7C-CF0F-4562-8943-F1F302134B07@newclarity.net>
 <f9e2f739-4964-b259-d9e5-4aeec938c54b@gmail.com>
 <F48FF1DC-5DE9-4C30-8BCF-C17A1877AF37@newclarity.net>
 <6f20bda1-eff6-631d-915f-d6bb149ee666@gmail.com>
 <93BA4B49-9D9A-4A41-83B3-8EC2564A70BF@newclarity.net>
 <CALKiJKoi7AH0ypkQN8p7S8jeO2M1qNnyMUfdDM+w4iLwhskskg@mail.gmail.com>
 <A819AFCE-875E-432E-9D87-EC4D970DF1DD@newclarity.net>
 <CAMuFt-6mw1s9fJ_Ndd2WzrFzDSNhjLdSm5x_iLxxyiB0AS5iCg@mail.gmail.com>
 <5455D4D6-9B07-4957-BA98-489EE12FEDD8@newclarity.net>
To: Robert Hickman <robehickman@gmail.com>
Subject: Re: [PHP-DEV] Initializing constants once, with code?
From: mike@newclarity.net (Mike Schinkel)

> On Jan 8, 2020, at 4:25 PM, Mike Schinkel <mike@newclarity.net> wrote:
>=20
>> On Jan 8, 2020, at 11:10 AM, Robert Hickman <robehickman@gmail.com> =
wrote:
>>=20
>> My suggestion was not a preprocessor, but what others have said
>> already, a 'shim' that goes between the parser/lexer and bytecode
>> generation which allows user code to inspect and modify the AST. This
>> idea is not unusual and Python for example can do this. The idea is
>> the basis of macro systems in Lisps, and Rust has a very nice
>> implementation of it from what I've seen in passing. The system =
should
>> be fully aware of the syntax, scoping, namespaces etc, the lack of
>> which is the problem with C preprocessor for example.
>>=20
>> With regards to how JAI does this:
>>=20
>> https://www.youtube.com/watch?v=3DUTqZNujQOlA
>> =
https://www.youtube.com/watch?v=3D59lKAlb6cRg&list=3DPLmV5I2fxaiCKfxMBrNsU=
1kgKJXD3PkyxO&index=3D20&t=3D0s
>=20
> Nice! =20

I replied prematurely. =20

OMG, JAI's ability to run code at _compile time_ is incredible, and =
exactly the kind of thing would solve the concerns about dynamic =
instantiation of constants.

It would also solve a large class of performance concerns, way beyond =
just dynamic instantiation of constants, as well as allow us to =
implement really robust error checking assertions into our libraries.

Which makes me wonder the following: AFAIK PHP opcodes are not something =
a developer can handle; they are only handled by PHP itself.

IOW =E2=80=94 correct me if I am wrong =E2=80=94 but PHP _cannot_ =
generate an OpCode file to disk that can be named and later loaded by a =
different PHP app?  The only thing similar to this is PHAR, which AFAIK =
is not OpCode, right?

If not, how about we discuss about creating a full API to allow =
developers to compile to OpCode files and then load and run them?  In =
addition to exploring this run PHP code to generate OpCode before the =
page load actually runs?

-Mike=