Nice. I think deleting code should be like hazard pay, double the normal hourly wage, but deleting Makefile lines should be triple wage. :) Looks great to me. Not that I use this makefile much anymore.
-kto On Aug 27, 2013, at 4:17 PM, Mike Duigou wrote: > Hello all; > > I have updated the changeset for this issue based upon feedback from the > earlier version. As a result of intervening work this version contains even > more cleanup. > > http://cr.openjdk.java.net/~mduigou/JDK-8015068/1/webrev/ > > Since the last revision: > > - One open issue remains--handling of shared_library_permissions. The > proposed patch uses a pattern rule, "jdk_%", rather than explicit make > targets. This means the knowledge of which targets required > shared_library_permissions has been lost. The shared_library_permissions > target could be run unconditionally as part of prep but I am uncertain if > this is reasonable. Alternatives? > > - The previously proposed behaviour of writing testoutput to the current > directory if not ALT_OUTPUTDIR is provided has been changed in response to > feedback. Output will now be written to jdk/testoutput and this directory has > been added to the .hgignore. Since most people are expected to use only the > root repo make which sets ALT_OUTPUTDIR it is not expected that this change > will make any difference for most users. > > As before, testing of this patch requires using a source build of JTReg as it > requires one fix that is not in the promoted builds. This changeset will not > be integrated until after the next JTReg promotion. > > Mike > > On Aug 7 2013, at 14:44 , Mike Duigou wrote: > >> Hello all; >> >> This changesest simplifies how the the jdk/test/Makefile processes excluded >> tests. Previously the test exclusions were pre-processed by scripts in the >> Makefile before being passed to JTReg. JTReg will now the all the >> processing. The change depends upon improvements in JTReg since the test >> exclude mechanism was originally defined. >> >> There are some changes in the reporting. Most obvious is that the >> excludelist.txt output file is no longer produced. The 'excluded' count in >> the summary report may be removed in a future reporting depending on whether >> the value can be calculated some other way. >> >> Additional to the exclusion list processing changes some other changes are >> possible as a result of the changes to exclusion processing. In particular >> the obsolete (and incorrect) logic for determining the build path included >> in the test/Makefile is now removed. If no output directory is provided via >> ALT_OUTPUTDIR then a default location in the current directory is used, >> "testoutput". At some point in the future this could be improved to get the >> CONF from the build configuration but that is not currently possible. >> >> Testing of this patch requires using a source build of JTReg as it requires >> one fix that is not in the promoted builds. >> >> http://cr.openjdk.java.net/~mduigou/JDK-8015068/0/webrev/ >> >> Mike >
