-
Bug
-
Resolution: Fixed
-
P3
-
26
-
master
Most of spec.gmk contains lines on the form:
VAR := @VAR@,
which are setup using AC_SUBST(VAR) in the configure script. However, there are also some other stuff there, which are mostly put there since we had no other good place to do so.
I have now created a new file, CommonVars.gmk, which is automatically included, and moved everything but pure configure assignments from spec.gmk into this file.
That makes spec.gmk a great interface between configure and make -- the job of configure is to fill in spec.gmk, plain and simple.
This refactoring also points to additional future improvements, e.g. generating spec.gmk automatically from all AC_SUBST statements, and to clean up the new CommonVars.gmk. Right now I kept all lines in the same order as they appeared in spec.gmk.template, to minimize risk of unintentional changes (we have some non-:= assignments), and to facilitate comparison for this patch. It is also clear that we have room for improvement in how we do bootcycle-builds and build the build-jdk. My current solution is to keep the special spec files, even if they just set a single special flag, to minimize impact of this patch.
VAR := @VAR@,
which are setup using AC_SUBST(VAR) in the configure script. However, there are also some other stuff there, which are mostly put there since we had no other good place to do so.
I have now created a new file, CommonVars.gmk, which is automatically included, and moved everything but pure configure assignments from spec.gmk into this file.
That makes spec.gmk a great interface between configure and make -- the job of configure is to fill in spec.gmk, plain and simple.
This refactoring also points to additional future improvements, e.g. generating spec.gmk automatically from all AC_SUBST statements, and to clean up the new CommonVars.gmk. Right now I kept all lines in the same order as they appeared in spec.gmk.template, to minimize risk of unintentional changes (we have some non-:= assignments), and to facilitate comparison for this patch. It is also clear that we have room for improvement in how we do bootcycle-builds and build the build-jdk. My current solution is to keep the special spec files, even if they just set a single special flag, to minimize impact of this patch.
- causes
-
JDK-8368714 [BACKOUT] JDK-8368468 Split out everything but configure results from spec.gmk
-
- Resolved
-
- links to
-
Commit(master) openjdk/jdk/de1f5a3c
-
Review(master) openjdk/jdk/27452