Just to be clear, I do not want to criticize Mark or the Project system in general. I quite *like* the project sequencing idea. I submitted the libada-gnattools-branch project as two phases, and I think any major changes from the second phase are perfectly suited to wait for stage 2. Furthermore, if there had been a major Ada or configury project lined up ahead of me, I would have been willing to delay phase 1 on the grounds of possible genuine merge conflicts (as opposed to the "phony" ones which make branch maintenance such a pain). I appreciate Mark Mitchell's work a great deal.
However, under the specific circumstances, I concluded that it was not appropriate to delay commiting the ready and tested first phase of the project. It also makes any future merges so much easier it's not funny. I do think certain types of janitorial projects should be given a higher priority for merging, because of how annoying they are to maintain on branches: specifically major rearrangement of code without intentional logic changes, especially between files. (This would be ameliorated for many cases by a switch to SVN, of course.) I found myself keeping *four* trees checked out routinely in order to get the code movement right when merging. -- This space intentionally left blank.