pcp
[Top] [All Lists]

Re: [pcp] pmrep: rename -R to -W

To: Nathan Scott <nathans@xxxxxxxxxx>
Subject: Re: [pcp] pmrep: rename -R to -W
From: Marko Myllynen <myllynen@xxxxxxxxxx>
Date: Mon, 14 Dec 2015 13:24:15 +0200
Cc: pcp developers <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <2007349003.40299535.1450067924129.JavaMail.zimbra@xxxxxxxxxx>
Organization: Red Hat
References: <566D93D8.9010109@xxxxxxxxxx> <2007349003.40299535.1450067924129.JavaMail.zimbra@xxxxxxxxxx>
Reply-to: myllynen@xxxxxxxxxx
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.4.0
Hi,

On 2015-12-14 06:38, Nathan Scott wrote:
> ----- Original Message -----
>>
>> related to the earlier raw/rate patches, the patch below renames -R to
>> -W in case we'd want to support requesting rates also for non-counters
>> in future (-W is not as good as -R for runtime but OTOH having -r/-R
>> available for raw/rate is nice).
> 
> Hmm, this looks alot like -T with a relative end point.  So, we should
> not need neither of these (-R / -W) options ...?

hmm, good point, I saw messages in the past time window options not
being supported with other than archives but I was probably using
something like -S or -T @10:00 at that time and mistakenly thought -T 2s
is not supported either. -R is used by pmcollectl and -T by pmnewlog for
this so -W is certainly not optimal.

> Perhaps the right thing to do here is to just remove this command
> line option at this stage, and implement -T more completely next
> release

Yeah, I have a look at this later today, I'll either send a two-liner to
drop -R and, if truly trivial, a patch to replace -R with -T.

> As we discussed elsewhere, we need to start getting regression test
> coverage of these patches too - anything this week will need to be
> regression tested, but in general it'd be good to see pmrep changes
> start arriving with test cases.

Agreed, but I haven't been able to run pmrep QA tests on my system so
far (as non-root). I would have hoped something like configure && make
&& make install && check 1069 would have worked but it seemed not to be
the case. Any guidance on this area would be helpful.

Thanks,

-- 
Marko Myllynen

<Prev in Thread] Current Thread [Next in Thread>