Re: $pipestatus broken?

[Frank Terbeck <ft@bewatermyfriend.org>]

Bart Schaefer wrote:
[...]
> With that loop, I get a single "1" 100% of the time unless I attach to
> the process with GDB, in which case it becomes random.


It's correct that this should output "0 0", though, right? With the right
hand side of the pipe being a loop, the last return code of the loop is
return code of the loop, correct?

If that's correct, the following adds a new test, that checks if the
problem happens on a given system (actually, if the problem appears at
least once in 2048 test runs - which has been triggered every time on my
test system so far). I didn't see anything about tests that are *expected*
to fail in `B01cd.ztst', so this makes the test suite fail at least once.
Therefore, I don't know if adding this is a good idea or not.

Regards, Frank


diff --git a/Test/A04redirect.ztst b/Test/A04redirect.ztst
index 3911b06..d349f33 100644
--- a/Test/A04redirect.ztst
+++ b/Test/A04redirect.ztst
@@ -441,3 +441,13 @@
 0:failed assignment on non-posix-special, POSIX_BUILTINS
 >output
 ?zsh: read-only variable: foo
+
+  repeat 2048; do (: | while read a; do
+                           :
+                       done;
+                       print "${pipestatus[@]}")
+  done | sort -u
+0:Check whether `$pipestatus[]' behaves.
+>0 0
+F:This test checks for a bug in `$pipestatus[]' handling.  If it breaks then
+F:the bug is still there or it reappeared. See workers-29973 for details.