accumulo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Keith Turner <ke...@deenlo.com>
Subject Re: mvn install stuck
Date Thu, 11 Apr 2013 18:01:32 GMT
On Thu, Apr 11, 2013 at 12:44 PM, Christopher <ctubbsii@apache.org> wrote:
> I can disable it by default... but it's just as easy to simply add
> -DskipLicenseCheck or to disable it for all your builds by adding the
> following to your settings.xml file:
>
>  <profiles>
>     <profile>
>       <id>deactivate-license-checks</id>
>       <properties>
>         <skipLicenseCheck>true</skipLicenseCheck>
>       </properties>
>     </profile>
>   </profiles>
>   <activeProfiles>
>     <activeProfile>deactivate-license-checks</activeProfile>
>   </activeProfiles>
>
> I understand the argument that it's configured by default for future
> changes, but it's not exactly true. It's configured by default to be
> pedantic, because current practices are a bit passive when it comes to
> these sorts of checks. This is useful for clean checkouts for
> releases, Jenkins builds, and for developers who don't run out of
> their workspace (they assemble first). I think the default should be
> pedantic, and it should take a minimal amount of effort to skip
> important checks, and that's something that we can immediately start
> benefiting from.

I agree that finding issue early is good.   If all developers  run out
of their workspace, wouldn't it be more convenient for everyone if it
were disabled by default in the pom?  I think the best counterargument
you have made for this is that it would be nice if the build server
would always do this check.   So is their an easy way to disable this
by default and have the build server pass in an option to turn it on?

This particular change does not really bother me. Its more the
precedent of changing the current build process to align with needed
improvements that will be made in the future.

>
> I'm still willing to make the change... if it is that much of a
> nuisance, but I emphatically argue against it for the above reasons.
>
>
> --
> Christopher L Tubbs II
> http://gravatar.com/ctubbsii
>
>
> On Thu, Apr 11, 2013 at 10:38 AM, Keith Turner <keith@deenlo.com> wrote:
>> On Wed, Apr 10, 2013 at 2:09 PM, Christopher <ctubbsii@apache.org> wrote:
>>> Both of those are addressed with the profile that is activated with
>>> -DskipLicenseCheck, so a dirty workspace will pass the check. The
>>> focus was on being pedantic for the clean checkout situation.
>>>
>>> We can add exceptions for those things that make a workspace dirty,
>>> but aren't packaged, for 1.5. However, in the future (>=1.6), I'd like
>>> to help make it easier to move away from the practice of dirtying the
>>> source directories to run Accumulo out of one's workspace.
>>>
>>> There is so much to maintain with all the svn:ignore properties set,
>>> the exceptions in the custom assembly descriptors and RPM/DEB
>>> profiles... it'd be better to allow running out of the target
>>> directory (which is already ignored by almost all Maven plugins), and
>>> use the default settings for packaging plugins wherever possible, than
>>> to worry about maintaining all these exceptions.
>>>
>>> Running out of the workspace can still be possible (out of the target
>>> directories, or a dedicated top-level workspace directory whose tree
>>> we ignore entirely), without all these exceptions to the rule.
>>>
>>> So, with that in mind, I only added exceptions to the apache-rat
>>> plugin configuration for things whose licenses are described elsewhere
>>> (js libs), or for things where it misinterprets the file as text
>>> instead of binary (splits, for testing), so that anything that was
>>> dirtying the workspace would explicitly be caught. As I said, it can
>>> be more lenient for 1.5 if you wish, but I think deactivating the
>>> check with the -DskipLicenseCheck should be sufficient for your needs.
>>
>> It sounds like the rat check is configured now for changes you plan to
>> make in the future.   IMO the poms should be configured for the way
>> developers work now.  The changes you mention sound great, when they
>> are made the rat check can be reconfigured.  Maybe that means
>> skipLicenseCheck is enabled by default for now, and has to be
>> explicitly disabled?  We do not want to put to much time into putting
>> bandaids on something that may fundamentally change, whats the
>> quickest solution to make it work?
>>
>>>
>>> I'm still curious, however, why things would have gotten stuck for
>>> you... getting stuck is very different than failing due to license
>>> checks.
>>>
>>> --
>>> Christopher L Tubbs II
>>> http://gravatar.com/ctubbsii
>>>
>>>
>>> On Wed, Apr 10, 2013 at 1:23 PM, John Vines <vines@apache.org> wrote:
>>>> Hmm, fresh checkout everything went fine. However, for sanity's sake I went
>>>> ahead and I dropped my configurations into conf and stripped out all of the
>>>> apache headers and I got a rat failure, too many unapproved licenses.It
>>>> shouldn't be checking those files since they aren't packaged.
>>>>
>>>> It also appears to be checking my log directory, so that needs to be
>>>> addressed too.
>>>>
>>>>
>>>> On Tue, Apr 9, 2013 at 8:05 PM, John Vines <vines@apache.org> wrote:
>>>>
>>>>> I had taken out the rat plugin in order to get it to build successfully.
I
>>>>> will try your tips tomorrow.
>>>>>
>>>>> Sent from my phone, please pardon the typos and brevity.
>>>>> On Apr 9, 2013 6:34 PM, "Christopher" <ctubbsii@apache.org> wrote:
>>>>>
>>>>>> That message is the from the apache-rat plugin, but the apache-rat
>>>>>> plugin would fail the build at verify phase if there was a problem.
It
>>>>>> wouldn't hang. You're going to have to provide more info, as it works
>>>>>> for me. Have you tried with a clean checkout? Does it work with
>>>>>> -DskipLicenseCheck? Does mvn package or mvn verify work?
>>>>>>
>>>>>> --
>>>>>> Christopher L Tubbs II
>>>>>> http://gravatar.com/ctubbsii
>>>>>>
>>>>>>
>>>>>> On Tue, Apr 9, 2013 at 5:44 PM, John Vines <vines@apache.org>
wrote:
>>>>>> > Attempting to do a mvn install of 1.5 and it just hangs with
the message
>>>>>> > [INFO] No excludes
>>>>>> >
>>>>>> > This has something to do with rat, but I don't know what.
>>>>>>
>>>>>

Mime
View raw message