udev bugfix

[Zbigniew Łuszpiński]

> Hello, zbiggy!
>
> On Thu, Jul 07, 2005 at 02:49:27PM +0200, Zbigniew ?uszpi?ski wrote:
> > I uploaded 2 patches to svn:
> > patch1 - original patch which I sent before
> > patch2 - patch which meets nestu's requirements (uses 49 than 50)
> >
> > Decide which one of these patches is better. The patch2 does some clean
> > up for people who installed my previous patch. Only one of these patches
> > can be used at once. Both does the same. And remember dead devices will
> > wake up after reboot.
> >
> > In my opinion patch1 is better than patch2 because it only adds hotplug
> > if there is none available. It do not looks dirty because it only adds
> > support for devices - nothing is removed.
>
> First of all I would like to thank you for your eagerness to help. It is
> very appreaciated ;)

??? I thought before I stick here and make devels nervous by messing svn with 
controversial, unwanted modules and unexpected patches (however very 
important IMO). But it is nice to know that people like to see some traffic 
on mail list and svn :-)

> In second place, I hate to say that the fix you are proposing still doesn't
> fit what we think should be done. As I said in my previous mail, we must
> not, under any circunstances, touch any other file that is not 49's rules.
> As you can see by my commit, we stick to this.

3rd attempt to bugfix:
Lunar by default install udev-061/etc/udev/udev.rules (as 50 file). How about 
using udev.rules from gentoo subdirectory instead? It contains all commands 
which my patch inserts into config files. And this file is original, 
unmodified, unchanged - provided with every udev release.

4th attempt to bugfix:
Build new lunar udev rules (including inside generic udev.rules) and bypass 
installation of generic udev.rules.

Temporary I run out of ideas, it looks nestu will have to show his secret 
weapon:
(...)
> apparently you also have a decent way of doing it. I look forward to 
> seeing that in moonbase :^P
Not really. That is a secret I will only keep to myself :þ
(...)

> BTW, if I hadn't been offline for a couple of hours due to an ISP problem,
> you would have seen the changes a bit earlier, since florin and I tested
> the fix this morning, about 7 hours ago. I did a little more work after, as
> I stripped the udev initscript too. Sorry about that.

Yes, I get used to see that some of devels like to strip almost naked... 
patches. :-D

Thanks for testing. I've just done lunar update, removed udev from zlocal and 
tried moonbase udev. Unfortunately I have no good messages. You applied 
extras part and 49 changes - that's good. But there is one thing. You didn't 
noticed that:
-Both tools must be executed from one file
-They can not be executed again from other rules file
Most code I added to BUILD script is written to reach such effect: place both 
commands in one file (49 or 50, no difference - thus 2 patches to choose) and 
clean up rest. The patches I send are not only tested by me 3x times but also 
stripped and optimized as much as possible. Thanks to this I never published 
broken patch. I do not make patches since yesterday, have some experience and 
know the patch should modify as less as possible. Thus I ask everyone to 
think twice before removing from my patches something more than credits (can 
not get used to this credit removal). I'd rather place credits inside patches 
for testers who can ask questions about code than for fun or pride.

The best thing is to borrow from someone any hotplug device which need 
firmware upload via kernel and test it with lunar. Then you will see why I 
made patches in this way not other.

I understand the need of keeping files original, not changed - especially 
those provided with external software which is not lunar specific. If it 
could be possible to do I would do that. But it doesn't in this case. Sorry. 
Ask udev creators to add hotplugd line at the end of udev.rules or use 
udev.rules from gentoo subdirectory.

> > greets,
> > Zbigniew 'zbiggy' Luszpinski
>
> Thanks again,

Thanks for partial implementation of patch and testing, finish is very 
close. :-)

Reverting zlocal/udev temporally (have to have working modem to access the 
Internet)

> nestu ;)

zbiggy :-)