IMHO, ordering of authors should not be taken into consideration: either you wrote the paper (and then you are an author: the paper would not be the same without you) or you did not (and then your name is not on it, maybe in the aknowledgments section if you participated undirectly); deciding who contributed the most is not relevant (what’s the most important? The number of lines written? The initial idea? The Developpement of the proofs? The experiments? Something else?). Mathematicians know how important it is to give author’s names in alphabetical order. I have always tried to follow that path, with exceptions (due to evaluations done by non-researchers of my collaborators)
Conference or Journal? In « old » sciences, journals are the only thing that matter. In Computer Science, it is the contrary: it is often harder to be published in great conferences than in great journals. Once you know that, most of the bibliometrics used by universities today are not relevant…
Impact Factor and other bù||$h!t$? Bibliometry is bad for science. It has the consequence to slow down progress and to put power into the hand of big editors who are making a lot of money on the work of researchers (see this paper and the citations it contains for details, for instance). In Computer Science, we do not want to use IF because our best journals and conferences have low IF. And then we have things like the CORE ranking: the French government uses it to evaluate French researchers in CS while it is made by some Australian private associations. As some fields of CS were not represented in Australia, this killed these fields in France. This literally implies that the French research policy about CS is decided in Australia…
The only thing that should really matter is open access. I mean real open access, where neither the authors nor the readers have to pay. We are doing the research, we are writing the papers, we are reviewing them, we are even editing them: why should we pay to access them? There exist a lot of places (conference proceedings and journals) where you can publish CS work that are open-access.