[Nut-upsdev] [nut-commits] svn commit r3071 - branches/apcsmart-dev/drivers

[Michal Soltys]

On 11-06-28 20:51, Arjen de Korte wrote:
> Citeren Michal Soltys <msoltyspl-guest at alioth.debian.org>:
>
>> Log:
>> apcsmart: allow tearing down/setting up serial connection during
>> normal activity;
>
> If this is needed at all (why?), I don't think this should be handled at
> driver level, but this would need to be in the serial.c library.
> Opening/closing a serial port is independent from the protocol used (no
> data is to be transmitted), so if there is a need for the apcsmart
> driver, likewise we will need this for other serial drivers as well. We
> certainly don't want to duplicate this code for every serial driver we
> support. One thing I fail to understand is why a driver should be able
> to close a connection to the serial port. Could you please elaborate on
> that?
>

It was one of the suggestions on that ubuntu's bugtracker (which Arnaud 
pointed me to). Supposedly reopening serial port could help with some 
pci based rs232 card. Flushing stale data was another thing to do (which 
the driver is doing now).

https://bugs.launchpad.net/ubuntu/+source/nut/+bug/535583/comments/10

Truth to be told, I don't really like the final effect - for the same 
reason you mentioned - as I pretty much duplicated the serial.c's code, 
with major difference being return <something> vs. fatalx() . The change 
is more along the lines "it won't hurt to do that just in case".

I could revert it to the earlier version and keep only non-serial 
related changes.

It works fine, but certainly feels out of place a bit.

>> changes should fix #535583 as well
>
> Please be a little more verbose in your patch comments. It took me a
> while to find that this was not a bug on Alioth, but on Ubuntu instead.
> This may be obvious for you, this information will be lost once this
> makes it into the drivers ChangeLog.
>
> Best regards, Arjen

Will do. I tried to be as terse as possible, though it looks like I 
overdid a bit.