[Splashy-devel] Bug#455685: ring around the testing

[Luca Capello]

Hello Luis!

Can you please *always* cc: your reply to the bug submitter (or the last
poster) if not otherwise stated?  You can do it directly or using the
NNN-submitter at b.d.o address [1].  This is the third bug I need to
manually check to know if there's any progress...

On Sun, 30 Dec 2007 12:59:16 +0100, Luis Mondesi wrote:
> The problem with this fix is that we do need to do something even if
> Splashy is not running. We need to figure a way to catch non 0 return
> codes after each step.

Sorry, I'm not a splashy expert, but I don't understand and I'd like to.

We can have two major boot situations when splashy is installed:

- vga=NNN splash

  fb is available and we need /etc/lsb-base-logging.sh

- only vga=NNN, only splash, nothing

  I fail to see why we need at all /etc/lsb-base-logging.sh in this
  case.  If for whatever reason splashy cannot start (and thus it's not
  running), why should we execute code related to it?  If this is the
  case, I'd suggest even a better patch (against the git repository):

--8<---------------cut here---------------start------------->8---
--- lsb-base-logging.sh.ORG	2008-01-05 12:41:10.000000000 +0100
+++ lsb-base-logging.sh	2008-01-05 15:48:41.000000000 +0100
@@ -31,6 +31,11 @@
     ##############################################################
     # Start splashy code 
 
+    # Bug #455685
+    # It makes no sense for us to send this step if splashy is not running
+    # Although then splashy_update would just return
+    pidof splashy > /dev/null || return $1
+
     # Bug #400598,#401999
     if [ -z "${RUNLEVEL:-}" ]; then
         # we need only the current level
@@ -49,10 +54,6 @@
     [ -x $SPL_UPD ] || return $1;
     [ -f $SPL_PRG ] || return $1; 
 
-    # It makes no sense for us to send this step if splashy is not running
-    # Although then splashy_update would just return
-    pidof splashy > /dev/null || return $1; 
-
     # Get progress percentage of this script
     # was calculated by update-progress-steps
     PER=`sed -n 's+'${0}' ++ p' $SPL_PRG`
@@ -113,12 +114,17 @@
     ##############################################################
     # Splashy code 
    
+    # Bug #455685
+    # It makes no sense for us to send this step if splashy is not running
+    # Although then splashy_update would just return
+    pidof splashy > /dev/null || return 0
+
     # send log to splashy. it will show if F2 is pressed
     pidof splashy > /dev/null && splashy_update "scroll $1: $2"
 
     # Stop splashy on *dm
     case $2 in 
-	?dm) stop_splashy || return 0 ;;
+	?dm) stop_splashy ;;
     esac
 }
 
--8<---------------cut here---------------end--------------->8---

The above patch has been tested on the four different combinations, but
only after having applied the fix from bug #459447 [2].

On Mon, 31 Dec 2007 06:35:54 +0100, Luis Mondesi wrote:
> I setup a test system with xdm and Splashy 0.3.9 [0] using a patched
> version of Directfb (I filed a bug [1] against libdirectfb-dev
> 1.0.1-5. This needs a patched that was already accepted upstream [2].
> Without this, Splashy takes 100% of the CPU while booting).

I think there's a missunderstand here: I'm in the situation where I've
no vga=NNN and no splash in /proc/cmdline, thus splashy is not running
at all.

> Can you please test this script? (all you need is to replace your
> /etc/lsb-base-logging.sh and reboot)

FWIW, the script works fine, but see my concerns above.

Thx, bye,
Gismo / Luca

Footnotes: 
[1] http://www.debian.org/Bugs/server-refcard
[2] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=459447