Skrypty startowe mldonkey

[Tomasz Pala]

On Thu, Dec 10, 2015 at 22:37:57 +0200, Elan Ruusamäe wrote:

> applied patch from rc-scripts mailing list dated 2015-11-26, thus three 
> weeks old!

So, you're saying there actually was not only a bug report, but the
patch itself and you did ignore it? Nice...

> if you're so smart and so caring, why didn't you apply it yourself. 

I'm not messing with rc-scripts so I don't read it's mailing list. If
you do, you should. Still not getting it?

> don't put pressure on me alone!

Please, try to understand - consider rc-scripts as STABLE release. No
new features, no improvements. Let it work like it did before. Commit
changes that are easy understandable, plain and simple, that CAN be
reviewed without knowing shell quirks. rc-scripts is used by less and
less people, so with PLD userbase each day it will be harder to find
such devious bugs.

> and if you had something to complain, why you didn't complain when the 
> commits were made?! package was sent builders week later.
> 
> and show me *your* super awesome commits.

Sure, here you are:

http://git.pld-linux.org/gitweb.cgi?p=projects/rc-scripts.git;a=commitdiff;h=00f5d898da7b7f0090519a6d727a6eb3d1e6dfb0

     introduced in rev. 8719 that neither pdksh nor mksh handled properly,
     not to mention bash - such process backgrounding caused, when called
     from functions/run_cmd (i.e. during service network start), I/O wait
     for STDOUT of 'background' job despite >/dev/null redirects; this
     effectively leads to 2 seconds delay for each interface at system
     startup, cumulating to minutes of pointless waiting on heavy networked
     machines! There is alternate workaround: { { sleep 2; } & } >/dev/null
     that works as expected under pdksh or mksh, but requires additional &
     for zsh and doesn't fix bash behaviour, so please do not touch _THIS_
     solution without comprehensive research! -- Aug 29th, 2012, /me
     Note: this happens due to $() capture, so plain ifup isn't enough.

   -               ( sleep 2; /sbin/arping -q -U -c 1 -I ${DEVICE} ${IP4ADDR%/*} ) > /dev/null 2>&1 < /dev/null &
   +               { ( sleep 2; /sbin/arping -q -U -c 1 -I ${DEVICE} ${IP4ADDR%/*} ) & } > /dev/null 2>&1

This is a comment quality required for this kind of changes. Not '2 seconds faster ifup'.
And a change itself - it makes ifup being much faster. And what did you do?

"rewrite $@ to modify command"
- that's obvious, but why? I would be against messing with positional/argument variables
  directly (only shift is a common practice)

"use runuser instead of su"
- why? did you upgrade util-linux R? No, you didn't... And this should be in a separate commit...

"this simplifies and avoids shell syntax. but until initlog, sigh"
- the hell, simplifies - and what's about that initlog?

> i've had enough of this, if you have bug to report, 
> http://bugs.pld-linux.org is the address where to submit them.

I don't need to fill a bug report, when I can revert bogus commits. I've
typed two suspected and they didn't introduce any bug solutions that
can't be removed.

-- 
Tomasz Pala <gotar at pld-linux.org>