Re: 5.9 core dumps under zleentry->zleread->zlecore

["Johan Ström" <johan@stromnet.se>]

On 2022-06-02 09:31, Bart Schaefer wrote:
> The end of the previous thread was that you had valgrind running, but
> we never heard any more from you after one report that was considered
> a false positive.
I think I submitted at least one valgrind post of initial warnings, 
which was probably bogus, but then I didn't really manage to reproduce 
it unfortunately, and it kind of got forgotten.
>
> We're going to need some more info on this, I fear -- would it at
> least be possible for you to configure --enable-zsh-debug and
> recompile, so that you get a stack trace with line numbers?

I've cloned 
https://github.com/archlinux/svntogit-packages/tree/packages/zsh/trunk 
now and added --enable-zsh-debug.
Running with that now, will report back if it crashes again.


During make check one failure occurred though, had to skip 'make check':

Testing multibyte with locale en_US.UTF-8
--- /tmp/zsh.ztst.4900/ztst.err    2022-06-02 11:12:41.488316802 +0200
+++ /tmp/zsh.ztst.4900/ztst.terr    2022-06-02 11:12:41.488316802 +0200
@@ -1 +1 @@
-./test_bad_param:1: command not found: $\M-i#
+./test_bad_param:1: invalid or incomplete multibyte or wide character: 
$\M-i#
Test ./D07multibyte.ztst failed: error output differs from expected as 
shown above for:
   # The following doesn't necessarily need UTF-8, but this gives
   # us the full effect --- if we parse this wrongly the \xe9
   # in combination with the tokenized input afterwards looks like a
   # valid UTF-8 character.  But it isn't.
   print $'$\xe9#``' >test_bad_param
   (setopt nonomatch
   . ./test_bad_param)
Was testing: Invalid parameter name with following tokenized input
./D07multibyte.ztst: test failed.