Oops DRILL-1548 :)

On Fri, Oct 17, 2014 at 5:39 PM, Hanifi Gunes <[email protected]> wrote:

> Let's track this on DRILL-1547 then.
>
> On Fri, Oct 17, 2014 at 4:43 PM, Aditya <[email protected]> wrote:
>
>> We don't.
>>
>> On Fri, Oct 17, 2014 at 4:05 PM, Hanifi Gunes <[email protected]>
>> wrote:
>>
>>> Agree that we should make sure nothing on jdbc side breaks in case we
>>> get take off jdbc-all from default build. Do we have a JIRA for this btw?
>>>
>>> -Hanifi
>>>
>>> On Tue, Oct 14, 2014 at 4:13 PM, Aditya <[email protected]> wrote:
>>>
>>>> I too, am inclined towards retaining the warning and removing
>>>> "jdbc-all" module from a non release build.
>>>>
>>>> However, a pre-commit test must activate this profile before checking
>>>> in a patch.
>>>>
>>>> On Tue, Oct 14, 2014 at 4:10 PM, Hanifi Gunes <[email protected]>
>>>> wrote:
>>>>
>>>>> I am inclined not using -dontwarn as it may mask possible issues that
>>>>> occur while proguarding. +1 for moving this from default build though.
>>>>>
>>>>> On Tue, Oct 14, 2014 at 4:05 PM, Aman Sinha <[email protected]>
>>>>> wrote:
>>>>>
>>>>>> Aditya,  thanks for those flags.  Do you want to create the JIRA for
>>>>>> removing it from the default build and perhaps assign it to yourself
>>>>>> :)
>>>>>>
>>>>>> On Tue, Oct 14, 2014 at 3:34 PM, Aditya <[email protected]>
>>>>>> wrote:
>>>>>>
>>>>>> > I agree. This could be made a part of "apache-release" profile.
>>>>>> >
>>>>>> > On Tue, Oct 14, 2014 at 3:32 PM, Jason Altekruse <
>>>>>> [email protected]
>>>>>> > >
>>>>>> > wrote:
>>>>>> >
>>>>>> > > I think the better option is to remove the proguard goal from the
>>>>>> default
>>>>>> > > build. It isn't testing anything or accomplishing a useful for
>>>>>> the dev
>>>>>> > team
>>>>>> > > to run it for every build, with or without the crazy logging. In
>>>>>> the
>>>>>> > > meantime this would be useful for making the build a little
>>>>>> faster.
>>>>>> > >
>>>>>> > > -Jason
>>>>>> > >
>>>>>> > > On Tue, Oct 14, 2014 at 3:16 PM, Aditya <[email protected]>
>>>>>> wrote:
>>>>>> > >
>>>>>> > >> This commit [1] turned off the verbose output from ProGuard
>>>>>> > >>
>>>>>> > >> However, the bulk of messages are info and warning, which
>>>>>> requires
>>>>>> > >> additional flags to turn off. The following patch can turn off
>>>>>> both of
>>>>>> > >> these
>>>>>> > >>
>>>>>> > >>  exec/jdbc-all/pom.xml | 2 ++
>>>>>> > >>  1 file changed, 2 insertions(+)
>>>>>> > >>
>>>>>> > >> diff --git a/exec/jdbc-all/pom.xml b/exec/jdbc-all/pom.xml
>>>>>> > >> index 349366b..783fa6f 100644
>>>>>> > >> --- a/exec/jdbc-all/pom.xml
>>>>>> > >> +++ b/exec/jdbc-all/pom.xml
>>>>>> > >> @@ -215,6 +215,8 @@
>>>>>> > >>
>>>>>> <outputDirectory>${project.build.directory}</outputDirectory>
>>>>>> > >>            <maxMemory>6g</maxMemory>
>>>>>> > >>            <options>
>>>>>> > >> +            <option>-dontnote</option>
>>>>>> > >> +            <option>-dontwarn</option>
>>>>>> > >>              <option>-dontobfuscate</option>
>>>>>> > >>              <option>-dontoptimize</option>
>>>>>> > >>              <option>-ignorewarnings</option>
>>>>>> > >>
>>>>>> > >> Do we want do do this?
>>>>>> > >>
>>>>>> > >> [1]
>>>>>> > >>
>>>>>> > >>
>>>>>> >
>>>>>> https://git-wip-us.apache.org/repos/asf?p=incubator-drill.git;a=commitdiff;h=35296501
>>>>>> > >>
>>>>>> > >> On Tue, Oct 14, 2014 at 2:32 PM, Hanifi Gunes <
>>>>>> [email protected]>
>>>>>> > >> wrote:
>>>>>> > >>
>>>>>> > >> > Afaik proguard plugin outputs verbose messages if it is a
>>>>>> debug build
>>>>>> > or
>>>>>> > >> > verbose switch is explicitly passed. Not sure about the recent
>>>>>> changes
>>>>>> > >> > though.
>>>>>> > >> >
>>>>>> > >> >
>>>>>> > >> > On Tue, Oct 14, 2014 at 2:11 PM, Jason Altekruse <
>>>>>> > >> [email protected]
>>>>>> > >> > >
>>>>>> > >> > wrote:
>>>>>> > >> >
>>>>>> > >> > > Not sure about the impact of the proguard upgrade, but I
>>>>>> almost
>>>>>> > always
>>>>>> > >> > > cancel the build if I happen to see that output. I know if
>>>>>> it gets
>>>>>> > to
>>>>>> > >> > that
>>>>>> > >> > > point that I haven't failed any tests. The only reason we
>>>>>> have that
>>>>>> > in
>>>>>> > >> > the
>>>>>> > >> > > build is to make the thin jdbc jar, which most devs on the
>>>>>> team
>>>>>> > aren't
>>>>>> > >> > > making use of right now. The longer term solution has been
>>>>>> > discussed,
>>>>>> > >> > which
>>>>>> > >> > > is pulling out the proguard step into a release only version
>>>>>> of the
>>>>>> > >> build
>>>>>> > >> > > that will be run less frequently.
>>>>>> > >> > >
>>>>>> > >> > > -Jason
>>>>>> > >> > >
>>>>>> > >> > > On Tue, Oct 14, 2014 at 1:58 PM, Aman Sinha <
>>>>>> [email protected]>
>>>>>> > >> wrote:
>>>>>> > >> > >
>>>>>> > >> > > > I am seeing a whole bunch of Proguard verbose output when
>>>>>> doing a
>>>>>> > >> build
>>>>>> > >> > > of
>>>>>> > >> > > > latest 0.7 master branch.   Anyone else seeing this ?  I
>>>>>> thought a
>>>>>> > >> > prior
>>>>>> > >> > > > commit had fixed this but maybe the upgrade of Proguard to
>>>>>> 5.0
>>>>>> > >> changed
>>>>>> > >> > > this
>>>>>> > >> > > > behavior.
>>>>>> > >> > > >
>>>>>> > >> > >
>>>>>> > >> >
>>>>>> > >>
>>>>>> > >
>>>>>> > >
>>>>>> >
>>>>>>
>>>>>
>>>>>
>>>>
>>>
>>
>

Reply via email to