udhcp missing prototypes

Rob Landley rob at landley.net
Thu Apr 20 00:01:01 UTC 2006


On Wednesday 19 April 2006 6:08 pm, Mike Frysinger wrote:
> On Wednesday 19 April 2006 17:11, Rob Landley wrote:
> > But the current udhcp code was doomed from the moment you deleted
> > busybox's copy of it.  When you deleted busybox's copy, it ceased to be
> > part of busybox.  Splicing in an external tree is irrelevant: that
> > doesn't ship.
>
> not really ... you were clearly all for killing udhcp before i moved things
> around ... my changes simply unified the code as it should have been

A quick grep of the patch list says svn 14634 is where you killed udhcp.

My busybox mbox has all my messages back to November 18, 2004, so tell kmail 
to sort by user, new folder "wurble", copy all my messages into there, go to 
the command line, and grep for udhcp...

December 6, 2004 mentioned dhcp in the subject, but now the message body.
Dec 11, 2004 I offered advice on diagnosing a dhcp related network problem.
Dec 18, 2004 I recommended against changing the then-current udhcp code...  
And the thread continues on a while.

Ok, skipping a bit...  May 24 2005 I followed up on a udhcp patch that I 
wasn't sure had ever been resolved/merged.  May 26, 2005 I checked in a fix 
to what I presume is the same issue.

Skipping a bit more, 3 more dhcp bug fixes went into 1.01.

August 21, 2005 I said:

> When you've got something for us to look at, send a standard diff -u patch 
to
> this list.  (The udhcp stuff is a bit weird because the maintainer of that
> applet also has it as an external standalone project, but keeping the two in
> sync is his problem...)

Skipping a few more cases where I discussed fixes to udhcp...

October 6, 2005:

> I avoid touching dhcp at all because then there would be some strange
> obligation put upon me to fix some out-of-tree version I actively don't care
> about.  I'm not going there.  It can rot for all I care.

Perhaps that's what you're talking about?  That was a declaration I was 
ignoring it, not removing it.  I didn't have the luxury of ignoring parts of 
busybox after I became maintainer.

More in passing mentions of or fixes to dhcp...

January 6 of this year, discussing the bug list for 1.1.0:

> 0000280     Networking Support  feature  [PATCH] Kill annoying log messages
> in udhcp-client
> 
> There's some sort of black magic I'm supposed to do to keep udhcp in sync
> with  the busybox version that's icky enough I'm avoiding touching it for
> now.  Revisit in 1.1.1...

Notice that I wasn't actually maintainer yet (I was still just filling in for 
Erik), so I still had the luxury of avoiding things, and that anybody else 
was quite welcome to fix that bug.  I had previously (December 17, 2005) 
indicated that bug 280 was on the "should fix" list, I just didn't get around 
to it and it was cut as not vital to shipping 1.1.0.

It came up again January 31, Feb 1, and Feb 13.  Feb 20 I acknowledged a fix 
to it (dunno if I checked it in or somebody else did).

Feb 25, I said:

> I never use ifup and ifdown because they need gratuitous configuration
> files. (This is the same reason I don't use the busybox udhcp client.  Just
> bring the darn interface up, don't make me write a script for you to call to
> do something obvious.)  I tried testing it just now, but apparently ubuntu
> doesn't have these config files, and ifconfig doesn't need 'em...

Did you take that as an indication that I planned to remove ifup and ifdown 
too?  (Hint: I don't.  There are people who use them, and I respect that.  I 
just don't happen to be one of them.)

On March 20 I was again involved in a conversation helping somebody debug 
udhcp, it was mentioned on march 23, again on march 23...

And then later on March 23 comes this message, my first response to you 
deleting busybox's dhcp directory (unhappiness):

http://busybox.net/lists/busybox/2006-March/019512.html

More unhappiness:
http://busybox.net/lists/busybox/2006-March/019514.html

Deep and abiding unhappiness:
http://busybox.net/lists/busybox/2006-March/019515.html

The first ultimatum that you ignored:
http://busybox.net/lists/busybox/2006-March/019521.html

My "ok, I'll bork the hell out of the external package by doing every possible 
way to tie it to busybox until you revert this mess" approach:
http://busybox.net/lists/busybox/2006-March/019598.html

(Which I couldn't do because my tree is borked.)

In this message I said:
http://busybox.net/lists/busybox/2006-March/019714.html

> I was happy to leave this thing alone indefinitely and focus on higher 
> priority things, until my hand was forced.  But I can't even build with the 
> tree spliced together like this, and even if I could it's obnoxious.  I 
> objected at the time.  For example, if I clean up the code to use 
> bb_getopt_ulflags() I break some unrelated project.  That's evil, and it's 
> NOT MY PROBLEM.
> 
> My objections were ignored, and now I'm making the problem _GO_AWAY_.  And
> the since only way to do that is write a new one, that's what I'm doing.

That was march 27, four days after I noticed it was deleted, and over three 
weeks ago now.  If that wasn't clear enough then, I don't know how I can be 
more clear now.

You say that I was "clearly all for killing udhcp before" you deleted 
busybox's copy of it.  Could you point me to where I gave you that impression 
in the archives?  Both the mailing list and irc have public archives, or if I 
sent you private mail on this particular topic this feel free to forward it 
to the list.  Apparently I miscommunicated and I'd like to figure out how.

For clarity, the "leave this thing alone" was that I was holding off on doing 
any unnecessary _cleanups_ to that code (like making it use 
getopt_ulflags()), because I was aware of the link to the external project 
and basically treating it like an infection I didn't want to poke at because 
it could only make things worse.  Notice, however, that it already didn't 
meet my needs, and the new one _will_.

I plan to run at least the new dhcp client I'm writing on my laptop as the 
default way it gets an address, so that if I plug into anything it can't work 
with I'll be motivated to _fix_ it.

> -mike

Rob
-- 
Never bet against the cheap plastic solution.



More information about the busybox mailing list