[Pm-utils] [PATCH 4/6] Do not overwrite the log, just append to it.

[Stefan Seyfried]

Richard Hughes wrote:

> I'm not sure about that. Victor and Michael basically rescued the
> project from ruins and built it back up.

I disagree with that.
"Ruins" would be something totally broken, which it wasn't. It was working
just fine. Ok, there was no support for userspace suspend, but all
distributions using userspace suspend just packaged up my patch for it and it
worked just fine.

Peter might not have been the best Project Leader wrt. responsiveness etc.,
but he had a clear plan for the project and a good design rule. As i
understood it, the design rule was "KISS" which is a good one for such a
almost trivial project.

> Of course, from ruins, you have
> to sometimes break API or internal structure to make things work in a
> maintainable way or to do something you couldn't do before.
> 
> If you care about legacy API for a long term support package, I suggest
> you patch the project from a spec file - which should be almost trivial
> to do.

I care about legacy API for the users: they learned how to write a hook and
where to put it and they learned how to configure the little bits. They don't
want to repeat that learning every year, especially not for such a trivial
piece of software.

Just read my rant again back when autoconf was introduced to this project -
this was as superfluous as most of the changes now, which are made just for
the sake of it. I am quite sure i said all that back then already.


>>>  If it is really a
>>> serious concern, patches to let us use syslog for logging (and get
>> log> rotation for free!) are welcome.
>>
>> I don't really care.
> 
> Then please don't make hurtful comments. You can't make an insult any
> less insulting by putting a ;-) at the end.

Using syslog does not make it better. This is a matter of known locations,
where people can find the log file. Of course i then can configure syslog to
write to pm-suspend.log, but as in RFC1925, "it is always possible to add
another level of indirection", this just makes no sense.

This project should be about users, that want to have an easy-to-understand
tool to make their machines suspend. Yes, it would be best if they would not
have to know anything about that tool, but i think we are a few years away
from that. So let's make that tool easy to understand. Let's make it easy to
write a hook that ejects that special device i have on my machine before
suspend. Let's make it easy to find the logfile. Let's make it easy to debug
failures.

What i am seeing here over the last months is exactly the opposite: it is
getting more complicated all the time.

While it may be a nice thing for a developer to have e.g. a plugin system for
suspend backends, the normal user, who is not a shell guru (and could not care
less about "bashisms") has no idea what is going on in that intertwingled hunk
of code.

This is exactly the opposite of the thing Peter wanted in the beginning, at
least that's what i believe.

And that's why i am resistant to pick up the changes or even package the new
pm-utils, because i don't want to break my users systems for no reason (and i
did not see anything working better when i tried it, so there is no reason)
other than "we change it, because we can".

> Sorry to be blunt.

Don't worry, i can handle that.

Have fun,

	Stefan
-- 
Stefan Seyfried
R&D Team Mobile Devices            |              "Any ideas, John?"
SUSE LINUX Products GmbH, Nürnberg | "Well, surrounding them's out."

This footer brought to you by insane German lawmakers:
SUSE Linux Products GmbH, GF: Markus Rex, HRB 16746 (AG Nürnberg)