[Lunar-commits] <moonbase-other> libgpod: Ah yeah. Need this so if systemd is installed its udev rules

[Auke Kok]

On 03/28/2013 10:11 AM, v4hn wrote:
> On Thu, Mar 28, 2013 at 11:10:28AM -0500, Dennis Veatch wrote:
>> On 03/28/2013 10:43 AM, v4hn wrote:
>>> please don't add any more "( ) > $C_FIFO 2>&1" to the moonbase.
>>> They are obsolete and just need to be stripped.
>>>
>>
>> [Wednesday, March 27, 2013] [1:54:23 PM] <dveatch> what is the
>> status of BUILDS regarding the use of (. ). &&, etc?
>> [Wednesday, March 27, 2013] [1:54:32 PM] <dveatch> are we at the
>> point of not using those anymore?
>> [Wednesday, March 27, 2013] [1:57:11 PM] <auke> no, they are still
>> needed in many cases
>> [Wednesday, March 27, 2013] [1:58:07 PM] <dveatch> then I interpret
>> your answer of cases they are not required
>> [Wednesday, March 27, 2013] [1:58:15 PM] <dveatch> case examples?
>> [Wednesday, March 27, 2013] [1:58:48 PM] <auke> so, I enabled -e and friends
>> [Wednesday, March 27, 2013] [1:59:02 PM] <auke> but in general it
>> doesn't work to catch-and-exit errors
>> [Wednesday, March 27, 2013] [1:59:05 PM] <auke> not sure why
>> [Wednesday, March 27, 2013] [1:59:14 PM] <auke> probably some bash
>> vodoo magic I missed
>
> you got that wrong, () > $C_FIFO _is_ obsolete and should be stripped.
> && on the other hand definitely __is not and should never be__ (imho).

that is the current state.

$C_FIFO redirection is automated, not a single BUILD can run without it 
being applied.

However, () blocks are something different. I thought that I could 
force-auto-enable this behavior, but that doesn't work properly, and so 
() and && blocks may still be needed to trap errors.

Auke