Thomas Gries wrote:
A certain enhancement patch Z is based on CVS TAG X
and
a split of Z into smaller patches p1, p2, p3 is requested,
Z := p1 + p2 + p3
What's your policy here, shall I publish
* diff( X, p1 )
* diff( X, p2 )
* diff( X, p3 )
or
* diff( X, p1 )
* diff( X + p1, p2),
* diff( X + p1 + p2, p3)
Probably the former. The purpose of multiple smaller patches is that one
of them (not necessarily the first) can be applied without the others.
Of course, the three patches should not conflict with each other.