[Ns-developers] cvs build warnings
tomh at tomh.org
Mon Sep 26 06:54:19 PDT 2005
>From: Mathieu Lacage [mailto:Mathieu.Lacage at sophia.inria.fr]
>Sent: Monday, September 26, 2005 07:35 AM
>Subject: Re: [Ns-developers] cvs build warnings
>On Thu, 2005-09-22 at 10:51 +0200, Mathieu Lacage wrote:
>> - changes the format of energy tracing. Although I don't care much
>> about this myself, it looks quite un-nice to do this to those who use
>> it. Or maybe there are no users ?
>I see that the reference output has been updated with this new energy
>tracing output. Is it really ok to do this ?
AFAIK this has been done on the project from time to time. There are a couple of concerns in doing this:
- is it truly a syntactical shift, or are behavioral changes masked by such an update? I think it has been the policy to trust the committer to have checked by hand that the output maps directly as a result of the change.
- are there third party users of the trace output whose software the change might break? For energy model, I don't know-- maybe not, but probably there are for general trace format. I would suggest that a backward compatibility conversion script to the old format be considered in that case. It would probably be a good gesture for the committer to probe the known heavy users about the impact of such a trace change, and discuss on the list.
>> I have checked in this patch to fix the compiler warnings. Please, can
>> the author of that patch take care of removing the dead code introduced
>> in cmu-trace.cc ?
>Is there anyone but me who cares about this ?
Padma, can you coordinate with Mathieu on this?
More information about the Ns-developers