Hannes, Thanks,
On the renamed message that you mention ... the old test code was broken, in a mildly funny way. Sorry for not mentioning it up front.
Initially, I changed the test as you might expect, and it started failing. On investigation, the test method being used takes a regular expression and so the final "." matched anything! So, the message being generated never changed, but the test is now more accurate when checking the message. Previously, it just checked an initial substring of the message.
-- Jon http://regex.info/blog/2006-09-15/247 On 6/9/20 6:47 AM, Hannes Wallnoefer wrote:
Looks good to me. I can’t quite work out why the message in testExplicitBinaryModuleOnLegacyPaths changes from module_not_found to module_not_found_on_sourcepath instead of module_source_not_found, but it seems like the correct message in the context. HannesAm 06.06.2020 um 01:48 schrieb Jonathan Gibbons <jonathan.gibb...@oracle.com>: Please review a small change to a poor warning message that is generated in unusual circumstances. The circumstances are caused by using --expand-requires and not having the source available for all the modules. • The message incorrectly uses the simple name of a module instead of the fully qualified name. The fix is trivial: use the fully qualified name • The message is too broad and says "module not found" when it really means "source for module not found" The message is improved • The message may be unnecessarily repeated. To minimize risk of a complicated solution, a simple cache is used to track instances of the warnings to avoid duplicates. In addition, a couple of inappropriate trailing periods are removed from a couple of related messages. Two tests are updated; one test is marked for this bug. -- Jon JBS: https://bugs.openjdk.java.net/browse/JDK-8246705 Webrev: http://cr.openjdk.java.net/~jjg/8246705/webrev.00/
