Failing test for 5.5: C03traps, W02jobs

Failing test for 5.5: C03traps, W02jobs

[Александр Капшуна]

Hello,

I was trying to build a brand new zsh 5.5 but `make test` finished with two failures. Several people also reported one or both of this fails.

Relevant parts of test run:


./C03traps.ztst: starting.
--- /tmp/zsh.ztst.20164/ztst.out        2018-04-10 18:20:58.850200661 +0300
+++ /tmp/zsh.ztst.20164/ztst.tout       2018-04-10 18:20:58.851200661 +0300
@@ -1,3 +1 @@
-trap -- int INT
-trap -- quit QUIT
-trap -- sigterm TERM
+trap -- quit 3
Test ./C03traps.ztst failed: output differs from expected as shown above for:
  trap -
  trap
  trap int INT
  trap sigterm SIGTERM
  trap quit 3
  trap
Error output:
(eval):trap:3: undefined signal: INT
(eval):trap:4: undefined signal: SIGTERM
Was testing:  Outputting traps correctly
./C03traps.ztst: test failed.

./W02jobs.ztst: starting.
Pattern match failed:
<\[1] [0-9]##
<\[2] [0-9]##
<\[3] [0-9]##
<\[4] [0-9]##
<\[4]  ? terminate*sleep*
<\[3]  ? hangup*sleep*
<\[2]  ? interrupt*sleep*
<\[1]  ? kill*sleep*
>[1] 24886
>[2] 24887
>[3] 24888
>[4] 24889
>[4]  +            sleep 7
>kill: unknown signal: SIGHUP
>kill: type kill -l for a list of signals
>kill: unknown signal: SIGINT
>kill: type kill -l for a list of signals
>kill: unknown signal: SIGKILL
>kill: type kill -l for a list of signals
>zsh: you have running jobs.
>zsh: warning: 3 jobs SIGHUPed
Test ./W02jobs.ztst failed: output differs from expected as shown above for:
  zpty_start
  zpty_input 'sleep 10 & sleep 9 & sleep 8 & sleep 7 &'
  sleep 0.1
  zpty_input 'kill %4'
  sleep 0.1
  zpty_input 'kill -HUP %3'
  sleep 0.1
  zpty_input 'kill -INT %2'
  sleep 0.1
  zpty_input 'kill -KILL %1'
  sleep 0.1
  zpty_stop
Was testing: various `kill` signals with multiple running jobs
./W02jobs.ztst: test failed.

And you can see full build log here https://paste.pound-python.org/show/3j0cgbBbq6vfxPmPrutE/ in case if you need it.

Asking for help in investigating issues with build environment or zsh itself. 

-- 
Alexander "kapsh" Kapshuna
kapsh@kap.sh

Re: Failing test for 5.5: C03traps, W02jobs

[Peter Stephenson]

On Tue, 10 Apr 2018 19:18:30 +0300
Александр Капшуна <kapsh@kap.sh> wrote:
> I was trying to build a brand new zsh 5.5 but `make test` finished
> with two failures. Several people also reported one or both of this
> fails.

Signal names not recognised --- I would guess this is down to the change
in glibc 2.25 reported (and fixed) by Benedikt Morbach in
zsh-workers/42618?

pws

Re: Failing test for 5.5: C03traps, W02jobs

[Alexander Kapshuna]

Yes, you are right. I am using glibc 2.27. 
Applying this patch fixes both of fails. Thanks for the help. 

./configure warns though about:

> checking where signal.h is located... ./configure: line 8998: test: too many arguments
>  /usr/x86_64-pc-linux-gnu/include/bits/signum.h /usr/x86_64-pc-linux-gnu/include/bits/signum-generic.h

But it seems to be harmless.


Can somebody also tell, is there any chance to see this fix in the close future? Something like hotfix release 5.5.1.

-- 
Alexander "kapsh" Kapshuna
kapsh@kap.sh


10.04.2018, 19:46, "Peter Stephenson" <p.stephenson@samsung.com>:
> On Tue, 10 Apr 2018 19:18:30 +0300
> Александр Капшуна <kapsh@kap.sh> wrote:
>>  I was trying to build a brand new zsh 5.5 but `make test` finished
>>  with two failures. Several people also reported one or both of this
>>  fails.
>
> Signal names not recognised --- I would guess this is down to the change
> in glibc 2.25 reported (and fixed) by Benedikt Morbach in
> zsh-workers/42618?
>
> pws

Re: Failing test for 5.5: C03traps, W02jobs

[Peter Stephenson]

On Tue, 10 Apr 2018 20:16:32 +0300
Alexander Kapshuna <kapsh@kap.sh> wrote:
> Yes, you are right. I am using glibc 2.27. 
> Applying this patch fixes both of fails. Thanks for the help. 
> 
> ./configure warns though about:
> 
> > checking where signal.h is located... ./configure: line 8998: test:
> > too many
> > arguments /usr/x86_64-pc-linux-gnu/include/bits/signum.h /usr/x86_64-pc-linux-gnu/include/bits/signum-generic.h  

Some extra quotes will help with that.

> Can somebody also tell, is there any chance to see this fix in the
> close future? Something like hotfix release 5.5.1.

Yes, I'll probably do that during the next week.

pws

diff --git a/configure.ac b/configure.ac
index c0686b6..d15a6cd 100644
--- a/configure.ac
+++ b/configure.ac
@@ -1553,12 +1553,12 @@ do
     SIGNAL_H="$SIGNAL_H $SIGNAL_TRY_H"
   fi
 done
-if test x$SIGNAL_H = x; then
+if test "x$SIGNAL_H" = x; then
   AC_MSG_ERROR(SIGNAL MACROS NOT FOUND:  please report to developers)
 fi
-zsh_cv_path_signal_h=$SIGNAL_H
+zsh_cv_path_signal_h="$SIGNAL_H"
 ])
-SIGNAL_H=$zsh_cv_path_signal_h
+SIGNAL_H="$zsh_cv_path_signal_h"
 AC_SUBST(SIGNAL_H)dnl
 
 dnl Where are error names located?  Needed as input for errnames1.awk

Re: Failing test for 5.5: C03traps, W02jobs

[Axel Beckert]

Hi,

On Tue, Apr 10, 2018 at 08:16:32PM +0300, Alexander Kapshuna wrote:
> Yes, you are right. I am using glibc 2.27. 
> Applying this patch fixes both of fails. Thanks for the help. 

JFTR: Builing on Debian Unstable with glibc 2.27 as package (which
implies passing the test suite) worked though fine. Didn't have to
patch anything.

On Debian Unstable, /usr/include/signal.h includes <bits/signum.h>
which is coming from /usr/include/bits/signum.h — might be a
debian-specific workaround for the same issue, though.

		Kind regards, Axel
-- 
PGP: 2FF9CD59612616B5      /~\  Plain Text Ribbon Campaign, http://arc.pasp.de/
Mail: abe@deuxchevaux.org  \ /  Say No to HTML in E-Mail and Usenet
Mail+Jabber: abe@noone.org  X
https://axel.beckert.ch/   / \  I love long mails: https://email.is-not-s.ms/