[Splashy-devel] Commit UNIX sockets

[Otavio Salvador]

Tim Dijkstra <newsuser at famdijkstra.org> writes:

> On Wed, 16 Aug 2006 18:21:10 -0300
> Otavio Salvador <otavio at debian.org> wrote:
>
>> Tim Dijkstra <newsuser at famdijkstra.org> writes:
>> 
>> > I also had a look at the start up script S01splashy, but I didn't had
>> > the stamina to fix it not starting splashy a second time;) So I just
>> > rudely made it not start splashy at all for testing (didn't commit it).
>> > If somebody wants to have a stab at that, please do. 
>> 
>> My idea is remove that huge code (init code) or write it on C.
>
> Yes, IMHO it tries to do to much things. For the case where splashy is run
> from initramfs it only has to start the client-daemon and that's that.

Well, yes and no. It currently does the math to see how many steps
will be need until the progressbar is complete. We'll need to do that
somewhere.

>> Most of init script are using lsb init functions when available so
>> might be easier to write one to provide this functionality instead of
>> all black magic done currently. What others think?
>
> I had the precise same idea! But I just grepped through /etc/init.d and
> there's still lots of scripts that do not use them. We could of course
> try to start a campaign to get it in to lintian/policy and submit
> patches...

Yes, indeed. But in that case we might have a way to handle the return
status as we do nowadays and use it to progress the bar. Just in case
we lack the lsb support for a script.

> The idea would be to depend on lsb-base and supply
> /etc/lsb-base-logging.sh which is designed to allow alternative
> implementations of the lsb log_* functions. We could then use the
> info we get via these functions to update the progress bar or
> display nice status messages. (I would have my splashy rather update
> a small text box, saying for example: Starting Network [OK]. And
> then, after clearing that line: Starting exim4 [ok], etc.) Much
> nicer than the progressbar IMHO.

That's the way usplash works nowadays but it's very simple and now so
flexible as splashy is and can be made. Nevertheless I think we can
use this information in verbose console mode instead of read tty
directly.

> Oh, something separate: the initramfs in the debian package checks for 
> vga in /proc/cmdline. This is broken, I'm using lilo and do not have vga
> on the kernel cmdline. 

Luis changed it yestarday. Hope it works better now.

-- 
        O T A V I O    S A L V A D O R
---------------------------------------------
 E-mail: otavio at debian.org      UIN: 5906116
 GNU/Linux User: 239058     GPG ID: 49A5F855
 Home Page: http://www.freedom.ind.br/otavio
---------------------------------------------
"Microsoft gives you Windows ... Linux gives
 you the whole house."