RFR: JDK-8033236 Update GensrcCharsetMapping.gmk to build-infra standards (original) (raw)

Xueming Shen xueming.shen at oracle.com
Mon Feb 3 23:11:33 UTC 2014

Changeset looks fine. Thanks for doing this.

Potentially it might worth considering to move two template files DoubleByte/SingleByte-X.java.template into the corresponding source code directory as well. But they can be dealt with later. We should migrate the StandardCharsets generating code from awk/sed to Java someday as well.

-Sherman

On 01/30/2014 06:29 AM, Alan Bateman wrote:

I think it's important that Sherman is at least one of the reviewers on this. -Alan. On 30/01/2014 14:18, Magnus Ihse Bursie wrote: The file GensrcCharsetMapping.gmk was not properly converted from the old build system.

This leads to unneccessary output, incorrect dependencies, missed chances for parallelization. It also breaks certain attempts to run static code analyzers such as Parfait. This change introduces the new API TextFileProcessing into jdk9. It's been in the build-infra project for a while. Similar to the compile constructs, we can setup a file (or a bunch of files) which needs different kinds of text transformations. In this case, only the INCLUDE functionality is used. This is also the first in hopefully a series of patches that will try to bring some order into the mess of generated/processed files. The following rules have been used: * Files that are almost java source files, but needs some text processing, should be named "Foo.java.template" and live in the correct place in src/classes. * Files that provides data to any gensrc action, but that are not at all java source code, should live in make/data/. Bug: https://bugs.openjdk.java.net/browse/JDK-8033236 WebRev: http://cr.openjdk.java.net/~ihse/JDK-8033236-update-gensrc-charset-mapping-gmk/webrev.01/ /Magnus

More information about the build-dev mailing list