The historical roles of journals used to be
The aforementioned change should effect the journal's policy. For example, in the past it was required that a paper that appears in a journal has not appeared in the same form (or version) anywhere else. This rule makes sense with respect to publication in other journals but not with respect to (prior) publication in a conference. It is indeed rare that a conference version is so complete and polished that it will not undergo a revision following careful review, but still the question remains (and the answer is clear: there is no scholarly reason to disallow journal publication of a version that is identical to a conference version, provided that this version is satisfactory for journal publication and that the relation to the conference version is stated (which is the norm anyhow)). (I am ignoring here copyright issues and refer only to the scholarly angle of this question. Furthermore, the scientific community should require that the publishers do not interfere in this matter, as a special case of requiring to preserve the right for free posting of its work in other non-commertial media.)
Similarly, I think that there is no justification anymore for having special issues devoted to a selection of papers from a conference that has a scope that matches the one of the journal. For further discussion of this issue see my opinion regarding STOC/FOCS special issues.
Back to Oded's page of essays and opinions or to Oded's homepage.