But Java would make it possible to use this "ancient codebase". And at the same time making support of many and new platforms easier.
Uh, how exactly would they go about doing that?
Java doesn't solve the problem of interoperability. It's not a magic bullet.
That Java is slower than native code is old myths. The overhead theese days is minimal and the speed of your disk will probably impact the speed of PS and Ligthroom an order of magnitude more.
Java proponents have been saying this for seven or eight years. It wasn't true that it was old myths then, it wasn't true two or three years ago, and I have sincere doubts as to it being a myth now.
Java proponents usually use examples of complex, well-written code being faster than complex, bad-written "native" code, just as the C or C++ proponents would use compact and elegant code compared to Java exception hell.
The overhead
is there yet, even though it's "minimal" and the speed of the disk always impacts the speed more. How much this impacts performance depends on your code, what your code does, etc. But do you see lots of technical computing applications being written in Java, or are the faster mathematical manipulations still done in C or FORTRAN?
But that's a sidetrack from the point I was making, which is that
PS would have been slower, if it had been implemented in Java already.
The reason for that is that if it had been implemented in Java already, it would most likely be using old libraries and interfaces for graphics handling anyway.
(Of course, you have to know that graphics handling in Java has improved a lot over the years in order to get that particular point, but still ...)