harmony-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vladimir Beliaev (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HARMONY-4587) [buildtest][eut] Expected Failure List for EUT3.3 on winx86
Date Mon, 06 Aug 2007 11:57:59 GMT

     [ https://issues.apache.org/jira/browse/HARMONY-4587?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Vladimir Beliaev updated HARMONY-4587:
--------------------------------------

    Attachment: HARMONY-4587-2-rename-EFL.sh

Corrected script - "touch" eut33.efl before adding it...

> [buildtest][eut] Expected Failure List for EUT3.3 on winx86
> -----------------------------------------------------------
>
>                 Key: HARMONY-4587
>                 URL: https://issues.apache.org/jira/browse/HARMONY-4587
>             Project: Harmony
>          Issue Type: Sub-task
>          Components: build - test - ci
>         Environment: Windows x86
>            Reporter: Vladimir Beliaev
>         Attachments: HARMONY-4587-1-redo-EFL.patch, HARMONY-4587-2-rename-EFL.sh, HARMONY-4587-2-rename-EFL.sh,
HARMONY-4587-3-eut33-efl.patch
>
>
> One need to create an Expected Failure List file for EUT3.3 on Windows x86. 
> It consist in several parts:
> 1. refactor tests\eut\build.xml to distinguish EUT3.2 and EUT3.3 EFL-files. A set of
eclipse bugs are fixed in Eclipse release 3.3, so EFL files for EUT3.2. and EUT3.3 will be
different. I believe there should NOT be an "unification" like "one EFL contains failure list
for both EUT versions". So it will be efl.eut32.* and efl.eut33 files set.
> Also It is a good point to create the "shared" EFL files for os / arch... Right now the
efl.linux.x86 = efl.windows.x86 and there will just few platform dependent failures after
linux failure analysis (for EUT32) is done. So, the proposed EFL "architecture" is:
>    eut32.efl - shared EFL file for all arch / os
>    eut32.windows.efl - windows specific failures (arch independent)
>    eut32.windows.x86.efl - windows x86 specific failures 
>    ...
> So while running on particular platform the EUT runner will concatenate shared and platform
specific EFL files and get thr resulted EFL file.
> 2. make EUT3.3 winx86 failure analysis and create eut33.windows.x86.efl file. This part
of task may be done just by putting all failed tests to this list with a comment "need to
do a failure analysis" w/o JIRA posting. This should be done so because:
>    - there is a set of failures (including intermittent ones) which are also reproducible
with RI Java (hot spot)
>    - there is a set of intermittent failures which are caused by "The GC did not provide
gc_add_weak_root_set_entry()" message 
>    - there is a set of failures with a minor contribution to pass rate (~ 0.1%)
> Thanks
> Vladimir Beliaev

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message