> In any event, the "nail in the coffin" for Bo's approach ended up being > security issues that were first raised by Andre. These can be hard to > foresee. It wasn't that these issues couldn't be addressed within Bo's > approach. It was, rather, that once you'd addressed them, you couldn't > have true "reversibility" any longer, and then the motivation for his > approach kind of evaporated, because that was the guiding idea.
While I will not continue our debate here, I would like to point out that, The reason why I chose to revert my feature was because no other developer agreed with the basic design of my implementation (i.e. in-place reversible bundle/unbundle), and I refused to change to a design that I dislike (i.e. force a directory structure, non-reversible bundle/unbundle). It was not because my implementation has any sort of fundamental flaw, as Richard indicated above. I have addressed Andre's security concern sufficiently, and I have proposed a method to achieve 'true reversibility' using my approach. I have communicated sufficiently with other developers, sometimes in private, why I left the development team. I would like to thank all lyx-devel watchers who expressed their opinions in private. I really appreciated your support. Have fun. Bo
