Re: Help testing release candidate / mc-4.8.26-rc1

[David Haller <dnh opensuse org>]

Hello,

On Sun, 10 Jan 2021, Yury V. Zaytsev wrote:
> TLDR; I would appreciate if you could please test the following tarball on
> your systems and report any blocker regressions as compared to the previous
> 4.8.25 release:

Two things:

1. mcedit causes a SIGSEGV when jumping out via Ctrl-o, even if just
   too _look_ at stuff, not doing anything on the (sub-)shell:

$ gdb /usr/bin/mcedit core.0-mcedit
[..]
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x00007f16a96eaca6 in do_update_prompt ()
(gdb) bt
#0  0x00007f16a96eaca6 in do_update_prompt ()
#1  0x00007f16a96eb2b5 in do_subshell_chdir ()
#2  0x00007f16a96eb4e7 in invoke_subshell ()
#3  0x00007f16a96666b1 in toggle_subshell ()
#4  0x00007f16a96e4061 in edit_dialog_command_execute ()
#5  0x00007f16a96e4813 in edit_dialog_callback ()
#6  0x00007f16a9670ad4 in dlg_process_event ()
#7  0x00007f16a9670e00 in dlg_run ()
#8  0x00007f16a96e3b00 in edit_files ()
#9  0x00007f16a9678717 in do_nc ()
#10 0x00007f16a9664505 in main ()

2. I don't know if that's a current regression or if that already was
   the case in 4.8.25 and earlier: if compiled without slang, no mouse
   actions work anymore (has the curses code gone?)

I can do more debugging if you can't pinpoint the change...

Ah, yes:

$ mc --version
GNU Midnight Commander 4.8.25-154-g33c84e75e
Built with GLib 2.64.5
Built with S-Lang 2.3.2 with terminfo database
With builtin Editor
With subshell support as default
With support for background operations
With mouse support on xterm and Linux console
With support for X11 events
With internationalization support
With multiple codepages support
With ext2fs attributes support
Virtual File Systems:
 cpiofs, tarfs, sfs, extfs, ext2undelfs, ftpfs, sftpfs
Data types:
 char: 8; int: 32; long: 64; void *: 64; size_t: 64; off_t: 64;

I have the detailed compilation log (and no relevant local patches
were applied, one just added some tmux/alacritty stuff to misc/mc.lib,
the other just changes doxygen stuff to generate pdf and manpages if
wanted).

Oh, and

3.: you should add the version to AC_INIT, so that PACKAGE_VERSION
    gets set correctly during configure. If not, generated docs have
    'mc-' instead of 'mc-$VERSION' (e.g. mc-4.8.25) as version in e.g. 
    the pdf ;)

A possible patch might be:

==== mc-4.8.26-fix_ac_version.patch ====
diff '-bpurNx*~' g/configure.ac h/configure.ac
--- g/configure.ac      2021-01-10 12:50:16.000000000 +0100
+++ h/configure.ac      2021-01-11 13:56:25.382961865 +0100
@@ -3,7 +3,8 @@ dnl Configure.in file for the Midnight C
 dnl
 
 AC_PREREQ(2.60)
-AC_INIT([GNU Midnight Commander], [], [mc-devel gnome org])
+m4_define(pkg_VERSION,[m4_esyscmd([sh ./ver.sh])])
+AC_INIT([GNU Midnight Commander], [pkg_VERSION], [mc-devel gnome org])
 m4_pattern_forbid(MC_)
 AC_CONFIG_MACRO_DIR([m4])
 AC_CONFIG_AUX_DIR(config)
diff '-bpurNx*~' g/ver.sh h/ver.sh
--- g/ver.sh    1970-01-01 01:00:00.000000000 +0100
+++ h/ver.sh    2021-01-11 13:54:30.000000000 +0100
@@ -0,0 +1,2 @@
+#!/bin/sh
+awk -F'"' '/MC_CURRENT_VERSION/{printf("%s",$2);}' version.h
====

(not using the external script get's you into quoting hell ;)

I've seen that it's somewhat redundant, but possible the later setting
of the version from version.h can be removed then, I've not looked
into that yet.

-dnh

-- 
My claim-to-fame in that respect involved a training-room-full of Suns
and a little script that would, at suitable intervals, cause one Sun to
"ping" in the style of a WWII Navy ASDIC echosounder, while another
would randomly respond with the appropriate echo-reply-sound.  -- Tanuki