Repository navigation
Functionality to allow set exotic ports names on Linux and MacOS - #182
andrey-patin-1979 wants to merge 5 commits into
Conversation
…cOS as was asked in (java-native#179 (comment)) and tests for this new functionality as was asked in (java-native#179 (comment))
|
I request we find a better word than |
|
Thanks for this effort. I believe some refactoring is required prior to merger. Thoughts welcome. |
So how do you want me to call them? |
|
@tresf, regarding error "/home/runner/work/jssc/jssc/src/test/java/jssc/SerialPortListTest.java:[27,63] method references are not supported in -source 6" |
|
Closing as non-planned, this is already covered by |
One test per mistake that reached users before, so each fails if it comes back (JOURNAL.md now maps lesson -> test): - java-native#113 baud rate truncated by a cast: set every standard rate and read it back with stty (verified: reintroducing a (jbyte) cast fails it) - java-native#192 0-byte reads on a real open port - java-native#86/java-native#181 event carries its port; removeEventListener true, then false - java-native#107/java-native#181 closePort after the device vanished does not throw - java-native#90 a port opened above fd 1024 still reads (poll on Linux) - java-native#80 the library loads under a Turkish default locale - java-native#160 exceptions stay IOExceptions - java-native#51/java-native#65 tests load the native library just built, not another copy - java-native#110 class files have the configured target version - java-native#179/java-native#180/java-native#182 Linux port names incl. Moxa, and getPortNames(Pattern) VirtualPortRule gets unplug() (stop socat) and one executor per rule: the shared static executor was shut down after the first test, so a class could only hold one socat test. Two tests found open bugs, recorded in BACKLOG: - reading from an unplugged port hangs on Linux because read() returns 0 after a hangup and the native loop retries (B-04); that test is @ignore'd until the fix - an incremental build without clean leaves Java 9 class files, because the second compiler pass sees up-to-date classes (new B-25) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RPFQMyWmP4EyGx33gCHP2r
One test per mistake that reached users before, so each fails if it comes back (JOURNAL.md now maps lesson -> test): - java-native#113 baud rate truncated by a cast: set every standard rate and read it back with stty (verified: reintroducing a (jbyte) cast fails it) - java-native#192 0-byte reads on a real open port - java-native#86/java-native#181 event carries its port; removeEventListener true, then false - java-native#107/java-native#181 closePort after the device vanished does not throw - java-native#90 a port opened above fd 1024 still reads (poll on Linux) - java-native#80 the library loads under a Turkish default locale - java-native#160 exceptions stay IOExceptions - java-native#51/java-native#65 tests load the native library just built, not another copy - java-native#110 class files have the configured target version - java-native#179/java-native#180/java-native#182 Linux port names incl. Moxa, and getPortNames(Pattern) VirtualPortRule gets unplug() (stop socat) and one executor per rule: the shared static executor was shut down after the first test, so a class could only hold one socat test. Two tests found open bugs, recorded in BACKLOG: - reading from an unplugged port hangs on Linux because read() returns 0 after a hangup and the native loop retries (B-04); that test is @ignore'd until the fix - an incremental build without clean leaves Java 9 class files, because the second compiler pass sees up-to-date classes (new B-25) Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01RPFQMyWmP4EyGx33gCHP2r
Added functionality for providing exotic port's names on Linux and MacOS as was asked in (#179 (comment)) and tests for this new functionality as was asked in (#179 (comment))