[Nut-upsuser] [Bug 278495] Re: megatec_usb does'nt detect ups

[Arnaud Quette]

Hi Alexander and Chuck,

2008/10/29  <lasaine at lvk.cs.msu.su>:
> Arnaud,
>
> On Wed, Oct 29, 2008 at 03:17:45PM +0100, Arnaud Quette wrote:
>> 2008/10/29  <lasaine at lvk.cs.msu.su>:
>> > Hi Arnaud,
>>
>> Hi Alexander,
>>
>> thanks for your sum up.
>>
>> > First of all I want to note that there is no phoenix subdriver in both
>> > the official 2.2.1 and 2.2.2 sources. If my memory serves me well I once
>> > sent a patch to Chuck that added it and some other stuff from the trunk.
>> > So the driver in 2.2.1 package was mostly the same as in the trunk
>> > because of this patch. It seems to me that something happened with this
>> > patch in 2.2.2 package (because 2.2.2 still lacks the new phoenix
>> > subdriver). So you see, it was actually a downgrade.
>> >
>> > The thing that confused me at first is that the 2.2.2 version log tells
>> > that the agiler subdriver is used. But this is just an error in log
>> > text. In fact it uses agiler_old subdriver which doesn't work with some
>> > models (like Artin's). Since this is already an official release, I
>> > can't fix it.
>>
>> right, I only checked the trunk. looking at a pristine 2.2.2, there is
>> only agiler + old and krauler
>> the trunk includes the phoenix sub driver.
>>
>
> Right
>
>> Note: just to clarify the situation, since it isn't clear to me (and
>> possibly others)
>> in the trunk, the 0x06da:0x0003 is listed with the phoenix_subdriver
>> agiler_old subdriver
>> in 2.2.2, it's listed with the agiler_old subdriver
>>
>> what is the right one and what is the right fix? This is the short run focus.
>>
>
> Well, I'll try to explain...
>
> Originally there was only the subdriver that is now known as agiler_old.
>
> The agiler subdriver is an Arjen's experiment. He thought that libusb
> will do all the buffer processing internally so this subdriver actually
> has one line of code in both get and set functions. But neither Arjen
> nor me couldn't test this code. This subdriver even made it into the
> 2.2.1 release and became the only subdriver from this "agiler family".
> However, the first tests showed that it doesn't work. After that Arjen
> together with the reporter wrote the phoenix subdriver which was
> intended as a replacement to the agiler_old subdriver.
>
> However by the time of 2.2.2 it was up to me to decide what to include
> to the release. The agiler_old driver was tested and stable so I decided
> to make it the default subdriver. The agiler subdriver was harmless so I
> leaved it untouched. But the phoenix subdriver was new and untested, and
> also I didn't like the code. So I didn't include it but made it the
> default in the trunk so that it could receive more testing.
>
> Now it proved to be superior to the agiler_old on some hardware (Artin's
> for example).
>
> When Ubuntu adopted 2.2.1 I had to send a patch to Chuck to make
> megatec_usb work again. This patch was really a diff from 2.2.1 to trunk
> so it included both the agiler_old and the phoenix subdrivers.

ok, the situation is now clear to me.
So the mentioned patch hasn't been reconducted for 2.2.2!
thanks for having recalled me everything (I'm managing far too many
things to recall everything ;-)

>> Also note that we can always extract a patch that only applies to (ie)
>> the Debian or Ubuntu package, without the need to throw out a new NUT
>> release...
>>
>
> I think, adding the phoenix subdriver to the Debian package would be ok
> too.

@Chuck: do you think we're still in time for a build?
@Alex: if Chuck approves, can you produce a quick patch?

thanks,
Arnaud