Marcus Denker <marcus at ira.uka.de> wrote:
[SNIP]
> Yes... somehow this starts to get unnececarily complicated: You are
> changing macrobenchmark methods, thus vm-generation removal
> needs to be filed in first. That's ok. 
> But my removal changes "discardVMRemoval" to not reference macroBenchmarks,
> which your removal actually allready deleted. (This it will be back after
> my "removal"). So I need to build a removal that requires the vm-removal
> to be files in first. Will do... 
> 
> After that I think I will stop doing any more removals until those that exist 
> get applied to the update stream... of course we could put a lot of work into
> the removals, but I think it would be *much* simpler to apply them as
> soon as possible (and then build new ones on top).

Right, perhaps we should simply start serious 3.6 development by
applying these to the update stream and then simply make sure those
removals aren't marked for "3.6". I think so.

> (Building Packages is nice, but the view that a Package allways lowers
> complexity is wrong: "Pakages" like removals or [ENH] of not-yet-modularised
> stuff are adding to the complexity, *not* lowering it. I'd even vote
> for adding "no-brainer" [ENH] to base-image-non-modularised stuff in the
> next release: Things like Yoshikis TrueType TextStyle, the "create Method"
> Debugger enhancement, and even Diegos look-improvements... the nice
> side-effect of this would be that people had actually a reason to upgrade:
> without improvements, not many will do it)

I would agree with the ENHs iff (note the two fs) they are "contained"
within a part of the image that eventually probably will turn into *one*
package. Point being that sure, ENHs are ok as long as they don't
increase the spaghetti metric. :-)


For example, improvements to Morphic (as those you listed) should
probably be allowed in with due review of course - because it will take
a while until Morphic is "packaged".

And I agree - it would make upgrading more interesting.

regards, Göran

PS. Doug and I will be trying to formulate the rough plan for 3.6 until
friday. Suggestions about it are much welcome and makes our job much
easier. And yes, it will only be a rough plan - nothing "carved in
stone".