Dani is correct that the survey was done because Gunnar agreed to implement the changes in EPP.
From: ide-dev-bounces@xxxxxxxxxxx [mailto:ide-dev-bounces@xxxxxxxxxxx] On Behalf Of Daniel Megert
Sent: December-09-13 11:07 PM
To: Discussions about the IDE
Subject: Re: [ide-dev] Survey results
Sorry to chime in late but the JDT team is currently very busy with (more) important work (Java 8) as Stephan already mentioned.
First, let me start to say that the survey rule was, that the survey needs to be backed by those who agree/want to make the changes. For this we/JDT never agreed to change the severity of all potential problems at once, but EPP offered to do this (I think). Having said that (as Stephan eloquently pointed out), it makes perfect sense to go over John's list and change some of them to 'Warning' in JDT, but others are just not OK to be changed for everyone. For those, I would also not recommend EPP to increase them, but that's up to the package owners.
Some Examples:
- 'Class overrides equals but not hashCode' makes sense for most users, hence I'd agree to enable this.
- There are many unused object allocations that are perfectly fine/desired, e.g. creating a new SWT Separator. Setting this to warning for everyone just sends the wrong message and there's currently no easy way to suppress this correct cases.
- 'Possible accidental boolean assignment e.g., if (a = b)' is a valid coding pattern. Some like it some think it is bad. Not us to judge.
- Boxing and unboxing conversions is not good to enable, because only one of them is unsafe. JDT should have separated those into two cases (see bug 163065 )
- At a first glance, missing default case probably also makes sense for most users.
As Stephan already mentioned, it's best to discuss this in detail in bug 423639, but of course EPP can move this forward if desired.
Dani
From: Stephan Herrmann <stephan.herrmann@xxxxxxxxx>
To: ide-dev@xxxxxxxxxxx
Date: 09.12.2013 22:32
Subject: Re: [ide-dev] Survey results
Sent by: ide-dev-bounces@xxxxxxxxxxx
To move this thread closer towards action I filed https://bugs.eclipse.org/423639
I will add a few considerations from JDT/Core pov in a minute.
I will be happy to work on this issue, but, as mentioned several times before,
right now is the most inappropriate of all moments for additional JDT work.
So, if doing this "after March and before April" is good enough I'll do my best.
For those who think that this is just about flipping a handful of switches:
I anticipate *significant* work to become necessary in order to adjust our
60,000+ tests to new compiler results. We must find a balance between keeping
our tests stable and testing the behavior that will be shipped as the new default.
Apart from that, experience tells that agreement about more aggressive warnings
is difficult to achieve among the JDT team, but keep in mind: if any new defaults
annoy our users, it will be our team that will be blamed, so please bear with some
detailed discussion, but let's move it to the bug, OK?
cheers,
Stephan
On 12/09/2013 03:45 PM, John Arthorne wrote:
> To be concrete, the survey talked about "Potential programming problems". In the Potential programming problems section, there are
> currently only the following set to ignore by default:
>
> Possible accidental boolean assignment e.g., if (a = b)
> Boxing and unboxing conversions
> Empty statement
> Unused object allocation
> Switch missing default case
> Switch case fall through
> Potential resource leak
> Missing synchronized modifier on inherited method
> Class overrides equals but not hashCode
>
> For the most part these are pretty reasonable warnings, with possible exception of box/unbox which is warning about simply using a
> feature of the Java language. Just be prepared that there is backlash *every* time we make a change that introduces warnings for
> people - although at least in this case we have some data to back up the decision.
>
> John
>
>
>
>
>
> From: Mickael Istria <mistria@xxxxxxxxxx>
> To: Ian Skerrett <ian.skerrett@xxxxxxxxxxx>, "'Discussions about the IDE'" <ide-dev@xxxxxxxxxxx>, "'Lars Vogel'"
> <lars.vogel@xxxxxxxxx>, "'Daniel Megert'" <daniel_megert@xxxxxxxxxx>,
> Date: 12/09/2013 09:13 AM
> Subject: Re: [ide-dev] Survey results
> Sent by: ide-dev-bounces@xxxxxxxxxxx
> ------------------------------------------------------------------------------------------------------------------------------------
>
>
>
> On 12/09/2013 02:53 PM, Ian Skerrett wrote:
> If you implement that change it will definitely annoy this minority.
> And what about the majority?
>
> Also, due to the nature of the change, my assumption is that a subset of people that said ‘yes’ did not appreciate the impact of the
> change so when it is implemented they will wish they had voted ‘no’.
> We've all reworked the question several times to make it explicit. I thought we've agreed the question was good enough so that we
> could trust the outcome of the survey and turn it into a concrete action (Yes or No to all warnings). Why deciding to almost ignore
> the vote now? Or why even asking the question if it's to ignore 65% of "Yes" ?
>
> 65% of people have expressed they'd like all warnings. We've discussed that the survey and reaction to results would also be a way
> to encourage the community to give feedback. I think letting 35% of community decide of everything is not fair at all.
> --
> Mickael Istria
> Eclipse developer at _JBoss, by Red Hat_ <http://www.jboss.org/tools>_
> __My blog_ <http://mickaelistria.wordpress.com/>- _My Tweets_
> <http://twitter.com/mickaelistria>_______________________________________________
> ide-dev mailing list
> ide-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/ide-dev
>
>
>
> _______________________________________________
> ide-dev mailing list
> ide-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/ide-dev
>
_______________________________________________
ide-dev mailing list
ide-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/ide-dev
