Hi Mike,
Won't pretend I follow all the changes, but how did you get rid of the
platform-specific checks in the ProblemList ? Will jtreg figure that
part out for itself?
Also what is the failure mode if JT_HOME is not set and jtreg is not in
the path? Should we just fail in that case before we try to use a blank
variable.
Thanks,
David
On 6/11/2013 4:13 AM, Mike Duigou wrote:
One final set of changes was required based upon cross platform testing.
The revised webrev which passes cross platform testing on all platforms is here:
http://cr.openjdk.java.net/~mduigou/JDK-8015068/3/webrev/
The changes are:
- I had to switch to using uname -s because Solaris doesn't support "-o" option.
- Some internal tools which scrape the results from Stats.txt were confused by the
"FIXME CODETOOLS-7900176" text. I have changed it to not include a space.
The second issue does point out one important difference between the previous
-exclude handling and the handling used by this changeset. The set of excludes
or summary counts is not part of the current logs. I would like to see the
excluded totals added into the runlist.txt file but currently jtreg doesn't
include either excluded or ignored in this list.
Are we ready yet? Finally?
Mike
On Oct 17 2013, at 05:23 , Alan Bateman <alan.bate...@oracle.com> wrote:
On 16/10/2013 21:22, Mike Duigou wrote:
Hello all;
With the imminent promotion of JTReg 4.1b07 it's possible to finally consider
completing this changeset!
http://cr.openjdk.java.net/~mduigou/JDK-8015068/2/webrev/
This updated webrev includes handling of the shared library permissions.
Thanks,
Mike
It would good to get this in when you can. I looked at the updated webrev and
it looks mostly the same as the previous rounds. The update to set the
permissions on DLLs looks fine, I assume that cygwin chmod must be add an ACE
that grants execute permission.
-Alan.
