Release Procedure

[AF]

It was mentioned that re-releasing etc was costly

What is the current release procedure beyond the meetings and the version number scheme, and what are the bottlenecks?

[SirMaverick]

Starting point: https://github.com/spring/spring/blob/m ... cklist.txt

[hoijui]

problem is, that there is always too little testing before, and too much bitching after, which makes it an angst thing.
the only way to solve it, is to have propper mutli version support, everywhere, by everythign and everyone, the same.

of course you could say, having a huge testing team bla bla... but i am talking about THE REALITY here.

and now licho will come and tell us that there is already propper multi version support, and koshi will say no ther is not, and...
it seems unskippable.

[Kloot]

There is never enough voluntary testing, but that wouldn't matter with a RERO policy (which was supposed to be the whole point of the current branching model, regardless of multi-version support...) whereas now the bitching about broken releases continues for weeks/months. Nobody would even need to care about user angst then either.

[hoijui]

the branching model was not supposed to address RERO, but to eliminate the ugly and tedious cherry picking.
RERO can only be achieved with smooth multi-version support everywhere.

[gajop]

I don't compile spring often, but when I do, I do it on arch linux!
Well, I can't compile it anymore without creating a patch for boost (using the one from ABS only works with stable releases), so I don't do it, hence why I can't be bothered to test it.
I used to be able to properly compile it before, what caused this change (changes in the boost & the float library you're using for development)?
I'd be happy to help testing, even perhaps fix some bugs (I think) I found, but it's currently a bit boring to do this.

But I'd rather write code that will do automated testing than spend my time manually testing it.