[hibernate-dev] "Stale" Hibernate issues
sanne at hibernate.org
Mon Apr 7 17:40:54 EDT 2014
Good points, we definitely need to make it easier to contribute, be it
just a least resistance path for making a test.
For what it's worth, when I started contributing more seriosly, the
"tipping point" which transformed me from an average user to a
fanatical contributor, was to understand how the testsuite worked. I
know it's not hard at all, but I was assuming that it was hard and so
I hadn't looked as I was assuming that I would not be in an easy
position to provide a working test. For example, I assumed I'd need
all the supported databases setup.
Let's not forget that most users don't ever start Hibernate directly
but get a Session via some framework, so this indeed requires some
I love Martijn's suggestion to upload an archetype with each release;
our first attempt could be very simple, then we take it from there and
maybe provide various mapped entities, a JPA and a native version...
On 7 April 2014 22:26, Brett Meyer <brmeyer at redhat.com> wrote:
> But Martijn, you're still missing my main points.
> "...awaiting a response from the core team..."
> "...without any evidence of a core member looking at the issue..."
> "...your users..."
> You're still thinking of this community in a proprietary product sense. It's not. The core team is small and focused. All individual contributors work on what they're most interested in or the biggest priority in their environment. If you have an issue that's important/urgent for you, then take a stab at fixing it! If everyone had that attitude within open source communities, some incredible things would result. Directly emailing the core team members, complaining that they haven't given your specific issue enough attention isn't going to accomplish anything. But that's certainly not to say that the core team is hands off. We all work hard. Really hard. And if something critical comes up, it's usually pounced on promptly.
> Before I started tackling the state of our JIRA instance, there were over 3k unresolved issues. Like I've already mentioned all over the place, that's not indicative of quality. It simply stems from a complex, 10+ year old project with thousands of users. When that many issues are open, it is *impossible* to identify what's still an issue. Going through them one-by-one is not even remotely reasonable. The policies we're trying to enforce are not simply to assist the core team, but the community as a whole. Imagine someone who is interested in getting started with contributing to ORM. If they pull up JIRA, how in the world would they know where to start?
> I will most definitely take a look at your project in HHH-9105 and roll something out for everyone to use as needed. Thanks for that. I'm definitely up for any more *constructive* ideas/feedback.
> Brett Meyer
> Red Hat, Hibernate ORM
> ----- Original Message -----
> From: "Martijn Dashorst" <martijn.dashorst at gmail.com>
> To: "Brett Meyer" <brmeyer at redhat.com>
> Cc: "Hibernate" <hibernate-dev at lists.jboss.org>
> Sent: Monday, April 7, 2014 4:29:38 PM
> Subject: Re: "Stale" Hibernate issues
> I know that open source projects are run by volunteers. I appreciate the
> efforts folks put in, even if they don't get to look at my issue-especially
> when I can work around the issue with a fugly solution, awaiting a response
> from the core team.
> I don't find my issue super duper important: we were able to work around
> it, and we check with every major release if someone has considered solving
> it-being disappointed, having a bitter laugh, and continue with our day to
> day job.
> My issue with the treatment is that after five years of neglect, without
> any evidence of a core member looking at the issue, it gets the "give us a
> test case or we will close it" treatment, combined with the "we don't want
> to look arrogant" attitude.
> There's a quid-pro-quo: if you ask your users to do something for you, you
> have to give something in return. As a user I have done quite a lot:
> described the issue I found in detail, provided mappings. Multiple users
> have reported the same behaviour, even in a 4.0-beta.
> Given that nobody from the core hibernate team has ever commented on the
> issue, I do remain with my opinion that the issue has not been taken
> seriously, giving me-a user who has put effort into reporting with proper
> detail what the issue is-the willies of ever reporting an issue with
> Hibernate again.
> So I have gone about and had to shave a yak or two to actually reproduce
> the issue at hand (creating a hibernate project, setting up a database,
> finding the magical incantations to start up a Hibernate configuration),
> fixing the logging, etc) taking up my free evening I could have spent on my
> own open source project, or socialising with my wife, or etc. All of this
> probably would have taken a core dev about 2 minutes given that you have
> the environment setup and know your testing framework in and out.
> So I will comment on HHH-3930 that the issue is still present in Hibernate
> 4.3.5, and also create a new issue with an empty test case which you can
> incorporate into your bug reporting template as a test case builder for
> folks reporting bugs. I hope you will incorporate this into your project
> and save everybody hours of time.
> See https://hibernate.atlassian.net/browse/HHH-9105 for the setup. It uses
> one deprecated API (config.buildSessionFactory()), but I figure you can
> improve upon this easily.
> This is also easily converted to a Maven archetype that can be generated
> with your normal release cycle. You can then create a online wizard that
> allows folks to quickly generate a quick start based on a specific version
> of Hibernate. See for example
> http://wicket.apache.org/start/quickstart.htmlfor such a page, and
> https://github.com/apache/wicket/tree/master/archetypes/quickstart for the
> implementation of that archetype.
> On Mon, Apr 7, 2014 at 9:05 PM, Brett Meyer <brmeyer at redhat.com> wrote:
>> Hi Martijn, I'm CC'ing the mailing list -- there are a few misconceptions
>> I'd like to clear up for everyone.
>> First and foremost, statements like "But nobody from the Hibernate has
>> ever taken a look" are frankly frustrating. Hibernate is an open source
>> project, not a proprietary product with separated groups of consumers and
>> producers. We all have our own interests and priorities. If HHH-3930 is
>> important for someone, pull requests or patches would be fantastic and much
>> appreciated by the community!
>> Your point argues against itself. There are so many issues that deserve
>> the community's attention, and HHH-3930 is no different from the rest. But
>> due to the sheer amount of open tickets, it was getting impossible to sort
>> through them all. We have to do *something* to narrow it down, and this
>> was the best solution we could find.
>> Everyone, please keep in mind that if an issue is still valid on ORM 4,
>> simply say so in a comment. I'm more than happy to re-open them on a
>> case-by-case basis. An aggressive approach was necessary -- simply letting
>> thousands of tickets sit there is, imho, much worse.
>> Regarding the development of test cases, we've said from the beginning
>> that, although we certainly prefer runnable cases (either standalone or
>> extending an existing unit test), I've always thought that enough detail in
>> the description (entities, mappings, applicable settings, and code
>> snippets) were perfectly acceptable and definitely making a big attempt at
>> helping out. That hasn't changed. However, several people have mentioned
>> wishing for a template project to use in creating standalone reproducers --
>> that's a great idea. I'm planning on putting one together soon and will
>> share it.
>> Brett Meyer
>> Red Hat, Hibernate ORM
>> ----- Original Message -----
>> From: "Martijn Dashorst" <martijn.dashorst at gmail.com>
>> To: brmeyer at redhat.com
>> Sent: Monday, April 7, 2014 2:41:20 PM
>> Subject: "Stale" Hibernate issues
>> Dear Brett,
>> I understand the necessity of sorting out stale issues in a large project
>> as Hibernate, but blindly shafting folks that are waiting 5 years for a
>> proper response from the hibernate team is not the way to go.
>> Case in point: HHH-3930 has been open for 5 years, has provided a clear
>> description and code for two entities afflicted entities. But nobody from
>> the Hibernate has ever taken a look and commented on it for specific
>> Only canned responses from mass updates have been provided.
>> I urge you to actually look at this issue and give proper feedback on what
>> is needed for it to be considered.
>> If the information is incomplete, or the report unclear, say so.
>> IMO you can only require folks adding a test case if you have a decent
>> guide of how to write one. Provide a maven archetype for it.
>> Requiring folks to have to checkout the hibernate project, figure out the
>> coding guidelines, having to peruse the test cases to figure out what to do
>> is really stretching it.
>> You state that you don't want to come across as arrogant, but from my
>> standpoint, the current actions actually make me wish I never had filed the
>> issue and just work around it, polluting my model and never going to file a
>> bug report with Hibernate ever again.
> Become a Wicket expert, learn from the best: http://wicketinaction.com
> hibernate-dev mailing list
> hibernate-dev at lists.jboss.org
More information about the hibernate-dev