You are viewing a plain text version of this content. The canonical link for it is here.
Posted to notifications@ant.apache.org by bu...@apache.org on 2019/10/22 18:51:38 UTC

[Bug 63874] New: Ant regression breaks javac openjfx build

https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

            Bug ID: 63874
           Summary: Ant regression breaks javac openjfx build
           Product: Ant
           Version: 1.10.7
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: regression
          Priority: P2
         Component: Core tasks
          Assignee: notifications@ant.apache.org
          Reporter: ilya.lipnitskiy@gmail.com
  Target Milestone: ---

Created attachment 36852
  --> https://bz.apache.org/bugzilla/attachment.cgi?id=36852&action=edit
Error log from ant 1.10.7

For some reason, javac command line argument files don't work with ant 1.10.7
when building Java openjfx. See attachment, or:
1. https://bugs.openjdk.java.net/browse/JDK-8231316
2. https://bugs.archlinux.org/task/64055

javac unexpectedly fails with:

[javac] error: invalid flag:
@/home/username/work/build/java-openjfx/repos/extra-x86_64/src/rt-13+14/build/compile.args

The cmd line argument file is well formed and the path is correct.

Downgrading to ant 1.10.6 fixes the problem.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

Jaikiran Pai <ja...@apache.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
   Target Milestone|---                         |1.10.8
         Resolution|---                         |FIXED

--- Comment #8 from Jaikiran Pai <ja...@apache.org> ---
The patch has been applied upstream. This fix will be available in 1.10.8 of
Ant. In the meantime, the Ant nightly build containing this fix can be used for
testing this change
https://builds.apache.org/job/Ant_Nightly/lastSuccessfulBuild/artifact/distribution/.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

--- Comment #2 from Jaikiran Pai <ja...@apache.org> ---
I remember there was a change in Ant, with code involving this. Let me see if
we caused some unexpected regression there.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

--- Comment #6 from Ilya Lipnitskiy <il...@gmail.com> ---
Also based on
https://docs.oracle.com/javase/7/docs/technotes/tools/windows/javac.html#commandlineargfile:
"Use of the '@' character to recursively interpret files is not supported." So
the whole changeset from https://github.com/apache/ant/pull/98 probably needs
to be reworked to make sure arguments that start with '@' don't get put into
commandline arg files.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

--- Comment #7 from Ilya Lipnitskiy <il...@gmail.com> ---
Should be fixed by https://github.com/apache/ant/pull/105!

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

--- Comment #3 from Jaikiran Pai <ja...@apache.org> ---
Is there a build.xml file to quickly reproduce this?

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

--- Comment #1 from Ilya Lipnitskiy <il...@gmail.com> ---
Created attachment 36853
  --> https://bz.apache.org/bugzilla/attachment.cgi?id=36853&action=edit
Success log with ant 1.10.6

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

--- Comment #4 from Ilya Lipnitskiy <il...@gmail.com> ---
Created attachment 36854
  --> https://bz.apache.org/bugzilla/attachment.cgi?id=36854&action=edit
build.xml excerpt

I don't have the time to minimize the build.xml at this point. However, if it
helps, the problematic invocation of javac is described by the attached
excerpt. The variables should be evident in the attached log.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

--- Comment #5 from Ilya Lipnitskiy <il...@gmail.com> ---
By examining the code in
src/main/org/apache/tools/ant/taskdefs/compilers/JavacExternal.java, it looks
like prior to commit 15fcc941e01, executeExternalCompile() was called with
cmd.size(), but in current code it is called with the index of the first non J
option after reshuffling the command line arguments, not cmd.size(). Seems like
a problem.

Moreover, the logic that looks at firstFileName in executeExternalCompile() is
only triggered when the length of the command like is greater than
COMMAND_LINE_LIMIT, which is 4KB on POSIX systems. Sounds like a test case is
needed for long command lines to cover this logic.

-- 
You are receiving this mail because:
You are the assignee for the bug.

[Bug 63874] Ant regression breaks javac openjfx build

Posted by bu...@apache.org.
https://bz.apache.org/bugzilla/show_bug.cgi?id=63874

Jaikiran Pai <ja...@apache.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |florian.bruckner@3kraft.com

--- Comment #9 from Jaikiran Pai <ja...@apache.org> ---
*** Bug 63111 has been marked as a duplicate of this bug. ***

-- 
You are receiving this mail because:
You are the assignee for the bug.