Alexander,
We probably need a clean separation of logging from error reporting as
Andy pointed out in his comment. I don't think having timestamps only in
verbose output is a good idea.
Besides at some point we might will use java.util.logging.Logger instead
of jdk.incubator.jpackage.internal.Log and having this clean separation
would make this transition easier.
- Alexey
On 7/7/2020 6:51 PM, alexander.matv...@oracle.com wrote:
Hi Andy,
Timestamps for error message without verbose output are meaningless in
my opinion. This is why I did not add them. Also, in some cases output
does not look right. For example when timestamp is added to error
message always:
jpackage --someoption
WARNING: Using incubator modules: jdk.incubator.jpackage
[15:49:23.798] Error: Invalid Option: [--someoption]
In example above I do not see any point to have timestamp.
Thanks,
Alexander
On 7/7/20 6:18 AM, Andy Herrick wrote:
All looks good, except maybe Log.error().
I think Log.error() should have the same output whether verbose or
not, adding timestamp to the message only if verbose does not look
right.
Problem is that Log.error() is used for two fundamentally different
purposes:
1.) by Main and Arguments to display the final error message when
jpackage is failing.
2.) by everyone else to display a serious warning message when
jpackage is continuing (sometimes failing further on, and sometimes
not).
I wouldn't mind adding timestamps only when verbose in case (2), but
I don't think that's appropriate for case (1).
/ANdy
On 7/6/2020 8:27 PM, alexander.matv...@oracle.com wrote:
Please review the jpackage fix for bug [1] at [2].
Added timestamp to verbose and test output in form of [HH:mm:ss.SSS].
[1] https://bugs.openjdk.java.net/browse/JDK-8248261
[2] http://cr.openjdk.java.net/~almatvee/8248261/webrev.00/
Thanks,
Alexander
