They released a bunch of fixes today.
One of them is easy to demo on VMS:
$ type another.php
<?php
$sslctxopts= [
-a-a-a 'ssl' => [
-a-a-a-a-a-a-a 'verify_peer'=> false,
-a-a-a-a-a-a-a 'verify_peer_name'=> false,
-a-a-a ],
];
echo file_get_contents("http://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; //echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx", false, stream_context_create($sslctxopts)) . "\n";
echo file_get_contents("HTTPS://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
echo file_get_contents("httpS://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
$ php another.phpTwo notes:
Port = 80
Port = 443
Port = 80
Port = 80
WTF????
reveal.aspx contains:
Port = <%= Request.Url.Port %>
echo file_get_contents("http://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; //echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx", false, stream_context_create($sslctxopts)) . "\n";And for the curious - before fix:
echo file_get_contents("HTTPS://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
echo file_get_contents("httpS://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
$ php another.php
Port = 80
Port = 443
Port = 80
Port = 80
On 11/19/2026 4:25 PM, Arne Vajhoj wrote:
echo file_get_contents("http://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
//echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; >> echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx", false,
stream_context_create($sslctxopts)) . "\n";
echo file_get_contents("HTTPS://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
echo file_get_contents("httpS://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
$ php another.phpAnd for the curious - before fix:
Port = 80
Port = 443
Port = 80
Port = 80
use_ssl = resource->scheme && (ZSTR_LEN(resource->scheme) > 4) && ZSTR_VAL(resource->scheme)[4] == 's';
On 2026-11-20, Arne Vajh|+j <arne@vajhoej.dk> wrote:Doesn't C require/mandate short circuiting in cases like this??
On 11/19/2026 4:25 PM, Arne Vajh|+j wrote:
echo file_get_contents("http://arnepc5.vajhoej.dk/reveal.aspx") . "\n";And for the curious - before fix:
//echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; >>> echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx", false,
stream_context_create($sslctxopts)) . "\n";
echo file_get_contents("HTTPS://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; >>> echo file_get_contents("httpS://arnepc5.vajhoej.dk/reveal.aspx") . "\n";
$ php another.php
Port = 80
Port = 443
Port = 80
Port = 80
use_ssl = resource->scheme && (ZSTR_LEN(resource->scheme) > 4) &&
ZSTR_VAL(resource->scheme)[4] == 's';
I really hope that line was always compiled by a short-circuiting
compiler and that line always successfully triggered short-circuiting...
Don't be surprised if everyone is now looking for similar constructs
that rely on short-circuiting to be safe.
On 11/20/2026 3:10 PM, Simon Clubley wrote:
On 2026-11-20, Arne Vajh|+j <arne@vajhoej.dk> wrote:
On 11/19/2026 4:25 PM, Arne Vajh|+j wrote:
echo file_get_contents("http://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; >>>> //echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; >>>> echo file_get_contents("https://arnepc5.vajhoej.dk/reveal.aspx", false, >>>> stream_context_create($sslctxopts)) . "\n";And for the curious - before fix:
echo file_get_contents("HTTPS://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; >>>> echo file_get_contents("httpS://arnepc5.vajhoej.dk/reveal.aspx") . "\n"; >>>
$ php another.php
Port = 80
Port = 443
Port = 80
Port = 80
use_ssl = resource->scheme && (ZSTR_LEN(resource->scheme) > 4) &&
ZSTR_VAL(resource->scheme)[4] == 's';
I really hope that line was always compiled by a short-circuiting
compiler and that line always successfully triggered short-circuiting...
Don't be surprised if everyone is now looking for similar constructs
that rely on short-circuiting to be safe.
Doesn't C require/mandate short circuiting in cases like this??
On 11/20/2026 3:10 PM, Simon Clubley wrote:I believe the VMS landscape looks like:
On 2026-11-20, Arne Vajh|+j <arne@vajhoej.dk> wrote:
And for the curious - before fix:
use_ssl = resource->scheme && (ZSTR_LEN(resource->scheme) > 4) &&
ZSTR_VAL(resource->scheme)[4] == 's';
I really hope that line was always compiled by a short-circuiting
compiler and that line always successfully triggered short-circuiting...
Don't be surprised if everyone is now looking for similar constructs
that rely on short-circuiting to be safe.
Doesn't C require/mandate short circuiting in cases like this??
PHP-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a &-a-a-a-a-a-a-a-a &&And in case somebody wonder.
-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a |-a-a-a-a-a-a-a-a ||
-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a and(**)
-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a-a or(**)
**) do not use ever
On 11/20/2026 3:24 PM, Arne Vajh|+j wrote:
On 11/20/2026 3:10 PM, Simon Clubley wrote:
[snip]Doesn't C require/mandate short circuiting in cases like this??
I believe the VMS landscape looks like:
Undefined(*) L to R L to R
all short circuit
[snip]
VMS Fortran .and.
.or.
VMS Pascal and and_then
or or_else
C & &&
| ||
Ada and and then
or or else
Java & &&
| ||
Python & and
| or
PHP & &&
| ||
and(**)
or(**)
[snip]
PHP & &&
| ||
and(**)
or(**)
**) do not use ever
In article <6b021bab$0$672$14726298@news.sunsite.dk>,
Arne Vajh|+j <arne@vajhoej.dk> wrote:
On 11/20/2026 3:24 PM, Arne Vajh|+j wrote:
On 11/20/2026 3:10 PM, Simon Clubley wrote:
[snip]Doesn't C require/mandate short circuiting in cases like this??
I believe the VMS landscape looks like:
Undefined(*) L to R L to R
all short circuit
[snip]
VMS Fortran .and.
.or.
VMS Pascal and and_then
or or_else
C & &&
| ||
Ada and and then
or or else
Java & &&
| ||
Python & and
| or
PHP & &&
| ||
and(**)
or(**)
[snip]
For C, Java, Python, and (probably?) PHP, you are mixing up
different categories of operators: `&` and `|` are bitwise
operators in those languages, while `&&` and `||` are boolean
operators. Since they are intended to do differnt things, the
semantics are different.
I doubt that there are many many languages that would introduce short-circuiting behavior for bitwise operators. But note alsoGood point.
that we may see behaviors like this:
```c
#include <stdio.h>
int
main(void)
{
int a = 1, b = 2;
if (a)
printf("a is true\n");
if (b)
printf("b is true\n");
if (a & b)
printf("both are true\n");
else
printf("but `&` is not a boolean operator\n");
return 0;
}
```
On Sat, 21 Nov 2026 17:32:01 -0500, Arne Vajh|+j wrote:
PHP & &&
| ||
and(**)
or(**)
**) do not use ever
How many things are there in PHP that one might classify as rCLdo not
use everrCY?
HererCOs another one: rCL+rCY on arrays. Yes, itrCOs defined. Defined as what?
Java has both bitwise & and | on integer types and logical & and |
on booleans.
In C and Python bool are sort of an int so bitwise and logical are
the same.
On Sat, 21 Nov 2026 20:25:26 -0500, Arne Vajh|+j wrote:
Java has both bitwise & and | on integer types and logical & and |
on booleans.
In C and Python bool are sort of an int so bitwise and logical are
the same.
Python treats bool as a subclass of int. (Yes, you can subclass int.)
Java tries to treat them as entirely separate types, which is not
necessarily a bad thing on its own,
but then it doesnrCOt provide
natural conversions between them. Like being able to use bool for
indexing arrays, just for example.
On 11/21/2026 6:58 PM, Dan Cross wrote:
In article <6b021bab$0$672$14726298@news.sunsite.dk>,
Arne Vajh|+j <arne@vajhoej.dk> wrote:
On 11/20/2026 3:24 PM, Arne Vajh|+j wrote:
On 11/20/2026 3:10 PM, Simon Clubley wrote:
[snip]Doesn't C require/mandate short circuiting in cases like this??
I believe the VMS landscape looks like:
Undefined(*) L to R L to R
all short circuit
[snip]
VMS Fortran .and.
.or.
VMS Pascal and and_then
or or_else
C & &&
| ||
Ada and and then
or or else
Java & &&
| ||
Python & and
| or
PHP & &&
| ||
and(**)
or(**)
[snip]
For C, Java, Python, and (probably?) PHP, you are mixing up
different categories of operators: `&` and `|` are bitwise
operators in those languages, while `&&` and `||` are boolean
operators. Since they are intended to do differnt things, the
semantics are different.
Those 4 are a bit different.
Java has both bitwise & and | on integer types and logical & and | on >booleans.
In C and Python bool are sort of an int so bitwise and logical
are the same.
PHP just auto converts booleans to ints for bitwise operator.
But all 4 quacks like a duck.
I doubt that there are many many languages that would introduce
short-circuiting behavior for bitwise operators. But note also
that we may see behaviors like this:
```c
#include <stdio.h>
int
main(void)
{
int a = 1, b = 2;
if (a)
printf("a is true\n");
if (b)
printf("b is true\n");
if (a & b)
printf("both are true\n");
else
printf("but `&` is not a boolean operator\n");
return 0;
}
```
Good point.
The problem here is the use of int instead of bool.
#include <stdio.h>
#include <stdbool.h>
int main()
{
bool a = 1, b = 2;
if (a)
printf("a is true\n");
if (b)
printf("b is true\n");
if (a & b)
printf("both are true\n");
else
printf("but `&` is not a boolean operator\n");
return 0;
}
Outputs a different result.
But treating int's as booleans is an integrated part
of C.
And it has spilled over to many other languages.
So yes - if working with int's instead of bool's,
then bitwise operators can not be used as logical
operators.
On Sat, 21 Nov 2026 17:32:01 -0500, Arne Vajh|+j wrote:
PHP & &&
| ||
and(**)
or(**)
**) do not use ever
How many things are there in PHP that one might classify as rCLdo not
use everrCY?
HererCOs another one: rCL+rCY on arrays. Yes, itrCOs defined. Defined as what?
| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 76 |
| Nodes: | 6 (1 / 5) |
| Uptime: | 35:14:38 |
| Calls: | 1,266 |
| Files: | 1,362 |
| D/L today: |
3 files (6,742K bytes) |
| Messages: | 284,775 |