On Aug 17, 2013, at 9:54 PM, Camillo Bruni <[email protected]> wrote:

> On 2013-08-17, at 21:26, Stéphane Ducasse <[email protected]> wrote:
>>> nope, the smalltalk hub diffs are clearly minor
>>> 
>>> - they are auto generate, no manual intervention needed to send them
>>> - they are only sent if the build succeeds
>>> - they include valuable information
>> like what?
> 
> I see exactly which classes and methods were touched with little effort.

I do not see the point to get a list of selectors.

> Which saves me clicks on each package and then browsing through the diff on 
> smalltalkhub.
This is ok now I do not think that we should continue to send these mails to 
the list because there are not diff
but list of selectors so they have much lower added values. 

I would be curious to know how many people systematically click on one link to 
see a change.
This is why I would like to avoid to force people to put filter on their 
mailing-list.  

>>> - the github repository is the place where you go to find changes, it's 
>>> simply impossible to figure out when something broke on smalltalkhub since 
>>> you cannot compare multiple packages over multiple times
>> 
>> I do not to go to github to see the diff because I will not do it and I want 
>> to see the code modified.
> 
> because?

Because this is what I want. I want to see the changes right in the mails in 
front of my nose.
It works well in Squeak so I do not see why it would not work for Pharo.


> 
>> It worked really well in the squeak mailing-list we did not do it because 
>> squeaksource was on its knees.
>> Nicolas will fix SmalltalkHub diff and I will see how the script loader can 
>> send the diffs
> 
> 
> I guess it's a matter of taste. I know that I need the git sources to 
> properly find bugs that were introduced by some faulty version.
> And usually I control each changes made to classes I maintain, which is very 
> easy to do for me when looking at the changes mail sent from github since I 
> see the touched methods on classes.
> A full diff, like what is done for VM-maker, is a bit too much for me, 
> especially when it comes in pure text form (compared to for instance what I 
> see on github?)

Probably true but this is another scenario. The scenario I want to see 
exercised is: more people look directly in the actual changes.

Stef




Reply via email to