Test failure on OpenBSD

Test failure on OpenBSD

[Matthew Martin]

[[ $'\ua0' =~ '^.$' ]]   exits non-zero on OpenBSD-current both in
5.2-test-2 and in 5.2. I swear the test was passing when I last updated
the port, but the test suite hasn't changed in the mean time.

Oddly on OpenBSD 5.9 and 6.0 with the packaged zsh 5.2,   echo $'\ya0'
errors with   zsh: character not in range

- Matthew Martin


./D07multibyte.ztst: starting.
Testing multibyte with locale en_US.UTF-8
--- /usr/ports/pobj/zsh-5.2-test-2/zsh-5.2-test-2/tmp.ztst.68282/ztst.out       Wed Dec  7 21:11:59 2016
+++ /usr/ports/pobj/zsh-5.2-test-2/zsh-5.2-test-2/tmp.ztst.68282/ztst.tout      Wed Dec  7 21:11:59 2016
@@ -1,3 +1,2 @@
 OK
 OK
-OK
Test ./D07multibyte.ztst failed: output differs from expected as shown above for:
  if zmodload zsh/regex 2>/dev/null; then
    [[ $'\ua0' =~ '^.$' ]] && print OK
    [[ $'\ua0' =~ $'^\ua0$' ]] && print OK
    [[ $'\ua0'X =~ '^X$' ]] || print OK
  else
    ZTST_skip="regexp library not found."
  fi
Was testing: Ensure no confusion on metafied input to regex module
./D07multibyte.ztst: test failed.

Re: Test failure on OpenBSD

[Peter Stephenson]

On Wed, 7 Dec 2016 21:26:52 -0600
Matthew Martin <phy1729@gmail.com> wrote:
> [[ $'\ua0' =~ '^.$' ]]   exits non-zero on OpenBSD-current both in
> 5.2-test-2 and in 5.2. I swear the test was passing when I last updated
> the port, but the test suite hasn't changed in the mean time.

I can't think of an easy way of looking at this from here.  Do you get
the same from an interactive command line, and with a completely clean
environment except for LANG=en_US.UTF-8 or similar?

That's fairly well through the tests so it's presumably specific to
what's going on in the regex library, suggesting some failure to set up
internationalisation properly down there.  Could some unhelpful
environment variable be leaking in (that's not looked at by the main
shell, so presumably not LC_ALL etc.)?

Preusmably also

[[ $'\ua0' == ? ]] && print OK

does work or the test would have bombed out before that point.

As it was already like this in 5.2 there's no point in holding up 5.3
(we can issue an update if it does get fixed any time soon), but some
research in OpenBSD regex set up is probably going to be useful.

pws

Re: Test failure on OpenBSD

[Matthew Martin]

On Thu, Dec 08, 2016 at 09:44:55AM +0000, Peter Stephenson wrote:
> On Wed, 7 Dec 2016 21:26:52 -0600
> Matthew Martin <phy1729@gmail.com> wrote:
> > [[ $'\ua0' =~ '^.$' ]]   exits non-zero on OpenBSD-current both in
> > 5.2-test-2 and in 5.2. I swear the test was passing when I last updated
> > the port, but the test suite hasn't changed in the mean time.
> 
> I can't think of an easy way of looking at this from here.  Do you get
> the same from an interactive command line, and with a completely clean
> environment except for LANG=en_US.UTF-8 or similar?

In   env -i LANG=en_US.UTF-8 zsh-5.2-test-2 -f   it still exits 1.

> That's fairly well through the tests so it's presumably specific to
> what's going on in the regex library, suggesting some failure to set up
> internationalisation properly down there.  Could some unhelpful
> environment variable be leaking in (that's not looked at by the main
> shell, so presumably not LC_ALL etc.)?
> 
> Preusmably also
> 
> [[ $'\ua0' == ? ]] && print OK

This does exit 0.

> does work or the test would have bombed out before that point.
> 
> As it was already like this in 5.2 there's no point in holding up 5.3
> (we can issue an update if it does get fixed any time soon), but some
> research in OpenBSD regex set up is probably going to be useful.

That's why I tested 5.2 as well. Any tips on where to look or what is
the C equivalent of    [[ $'\ua0' =~ '^.$' ]] ?

Re: Test failure on OpenBSD

[Bart Schaefer]

[-- Attachment #1: Type: text/plain, Size: 584 bytes --]

On Thu, Dec 8, 2016 at 9:13 PM, Matthew Martin <phy1729@gmail.com> wrote:
>
> That's why I tested 5.2 as well. Any tips on where to look or what is
> the C equivalent of    [[ $'\ua0' =~ '^.$' ]] ?

According to http://man.openbsd.org/regex.3 --

 There is one known functionality bug. The implementation of
internationalization
 is incomplete: the locale is always assumed to be the default one of IEEE
Std
 1003.2 (“POSIX.2”), and only the collating elements etc. of that locale
are available.

Wouldn't that mean this test should be expected to fail on openbsd?

Re: Test failure on OpenBSD

[Peter Stephenson]

On Fri, 9 Dec 2016 00:24:33 -0800
Bart Schaefer <schaefer@brasslantern.com> wrote:
> According to http://man.openbsd.org/regex.3 --
> 
>  There is one known functionality bug. The implementation of
> internationalization
>  is incomplete: the locale is always assumed to be the default one of IEEE
> Std
>  1003.2 (“POSIX.2”), and only the collating elements etc. of that locale
> are available.
> 
> Wouldn't that mean this test should be expected to fail on openbsd?

diff --git a/Test/D07multibyte.ztst b/Test/D07multibyte.ztst
index 3a6e955..9a246ca 100644
--- a/Test/D07multibyte.ztst
+++ b/Test/D07multibyte.ztst
@@ -511,6 +511,8 @@
 >OK
 >OK
 >OK
+F:A failure here may indicate the system regex library does not
+F:support character sets outside the portable 7-bit range.
 
   () {
      emulate -L zsh

pws

Re: Test failure on OpenBSD

[Peter Stephenson]

Second thoughts again... there isn't much external dependence in D07
apart from standard parts of internationali[sz]ation like locale, so it
might make sense to move this test last and execute everything related
to zsh's internals and the C library first.

pws

diff --git a/Test/D07multibyte.ztst b/Test/D07multibyte.ztst
index 3a6e955..cdc070e 100644
--- a/Test/D07multibyte.ztst
+++ b/Test/D07multibyte.ztst
@@ -500,18 +500,6 @@
 # aren't quite double width, but the arithmetic is correct.
 # It appears just to be an effect of the font.
 
-  if zmodload zsh/regex 2>/dev/null; then
-    [[ $'\ua0' =~ '^.$' ]] && print OK
-    [[ $'\ua0' =~ $'^\ua0$' ]] && print OK
-    [[ $'\ua0'X =~ '^X$' ]] || print OK
-  else
-    ZTST_skip="regexp library not found."
-  fi
-0:Ensure no confusion on metafied input to regex module
->OK
->OK
->OK
-
   () {
      emulate -L zsh
      setopt errreturn
@@ -583,3 +571,19 @@
   printf '%q%q\n' 你你
 0:printf %q and quotestring and general metafy / token madness
 >你你
+
+# This test is kept last as it introduces an additional
+# dependency on the system regex library.
+  if zmodload zsh/regex 2>/dev/null; then
+    [[ $'\ua0' =~ '^.$' ]] && print OK
+    [[ $'\ua0' =~ $'^\ua0$' ]] && print OK
+    [[ $'\ua0'X =~ '^X$' ]] || print OK
+  else
+    ZTST_skip="regexp library not found."
+  fi
+0:Ensure no confusion on metafied input to regex module
+>OK
+>OK
+>OK
+F:A failure here may indicate the system regex library does not
+F:support character sets outside the portable 7-bit range.