Keith Hodges wrote:
>> This counter example, code languishing in the Testing repository since
>> 2006 had failed to be integrated, even though it has been loadable from
>> Universes for most of that time, and was positioned as the new head for
>> everyone to work from. If people were to use it as the new head then
>> there wouldn't be any integration needed at all.
>
> That's a strange sense of entitlement you have here. Why do you think
> anyone would heed your "positioning as the new head for everyone to
> work from"?
1) Because it is the only one in a public repository where contributions
are invited, so by default it is the only one in the ownership of the
community.  If there was any other one, then I would have contributed to it.

2) It is the only one that has had significant work done on it in the
past 3 years, so it is therefore the only one that has moved forward,
and so by default it is the only head. If you pick the SUnit in 3.9 then
what you are picking is more of a shoulder or navel.

3) Because it is the only one that attempts to address the articulated
goals of the communities using SUnit. So really you need to rephrase the
question with the broader context in mind.
> This is but one of the many versions of SUnit and I'm not sure what
> would make it so special. In fact, given that it's not used in *any* fork 
In this public form it post dates most recent developments except for pharo.
> today it seems especially peculiar that you seem to be claiming it to
> be base of development for *all* of them.
Precisely because it is the only one maintained as an external package,
makes it most likely to be the only one that is capable of fulfilling
the role.

Since a number of forks have the professed goal of automated testing,
then the TestReporter is only present in this version.
>> My point is that by putting the philosophy first, that is what leads you
>> to have tools like MC for interchanging code in the first place, and it
>> is what led me to spend time working on MC1.5. SUnit falls in to the
>> same category of tools that essentially have to be common across forks.
>>
>> But if no one else adopts even a similar philosophy what's the point.
> I'm not sure what your philosophy is. 
To embrace the contributions of the community. To enable people to both
diversify and consolidate. Make it easy to fork (i.e. build an image to
your tastes) and easy to share fundamental common parts.
> If it is "let's make sure we use the same SUnit version across
> different forks" then you should rejoice: As of today, Squeak uses the
> same SUnit and SUnitGUI as Pharo! 
Which is no use to me, since there is no TestReporter in that version.
> We just reduced the burden of integration for anyone who has
> dependencies on either of the two.
Ok some challenges for you.

1) now try and write a single test suite for both Pharo and Squeak that
reports all tests green for each image, even though there may be
significant differences.

2) I want to take an image and test it from the command line. The
outputs I want are a progress indicator, and files with failures and errors.

3) I want to run SSpec specifications in the same runner (not functional
yet, but much groundwork has been done).
etc etc.

This version is the only version that has been developeed with the
professed goals of Pharo and Squeak in mind. If you dont have those
goals then do what you like,  but we do have those goals (well we used to)

Keith