[Rivet] [Rivet-svn] r3169 - trunk/bin

[Frank Siegert]

On 29/06/11 23:04, Andy Buckley wrote:
> On 29/06/11 21:54, Frank Siegert wrote:
>> On 29/06/11 22:50, Andy Buckley wrote:
>>> Well, my #1 reason is the principle of least surprise: it's IMO the
>>> "normal" behaviour that the last value that a parameter is set to is the
>>> one that applies. In fact, I have a lot of existing .dat files which
>>> have specifically provided an override for some generated value which
>>> presumably will have stopped working in r3139! I only discovered this
>>> behaviour when being confused by some .dat files with a config that used
>>> to work!
>>
>> But in the case where you have an override in a .dat file you wouldn't
>> have yet another .dat-file-specific override in your make-plots.conf
>> which you want to be ignored, would you?
>> This is the use case that I was talking about... if the implementation
>> in make-plots did not (only) do that, then I agree that we should look
>> at it again.
>
> Yeah, I had a feeling there was going to be something like that going on ;)
>
> The implementation that I reversed this evening meant that if a header
> section contained
>
> XMin=0
> Xmin=1
>
> (for example), then the effective value of XMin when drawing the plots
> would be 0. I don't think we want that!

This should hopefully be fixed with r3264, i.e. both cases described 
above work now.

Cheers,
Frank