As soon as people change any of the default compiler settings on a per project level, JDT will store all of the preferences in a file called 'org.eclipse.jdt.core.prefs'. So this file will include the non-changed values as well. You cannot tell whether it was an explicit decision if the value matches the default, but you know that at least it's a user who knows about per project preferences and cares changing and maintaining them.
So these users have definitely touched *some* defaults.
I think a good mix of such data, survey information and common sense will help to find which compiler checks should be enabled. I totally agree that there are some which should be turned on by default.
Good point Lars...
Sven, when this search runs does it return values that are not explicitly *in* the .project file or does it always return a result for a given preference ? If not (i.e. the total of the setting values always == the total number of .project files) then is there any way to determine whether the value was explicit or not ?
What I'm trying to do is to reduce any 'default' bias by only scraping preferences that have been explicitly over-ridden. If we can do this then the results would be even more compelling since they'd represent those values that someone has found a real need to change from the default (or to explicitly enforce).
BTW, just for fun...;-) What do we get when we look at the values for 'show line numbers' ?
Back to testing,
Eric
<graycol.gif>Lars Vogel ---12/10/2013 04:39:32 AM---Hi Sven, IMHO most users don't touch defaults, this might be the reason why you find
Hi Sven,
IMHO most users don't touch defaults, this might be the reason why you find the default setting so often on Github.
Best regards, Lars
2013/12/10 Sven Efftinge <sven@xxxxxxxxxxx>
On Dec 9, 2013, at 5:46 PM, Mickael Istria <mistria@xxxxxxxxxx> wrote:
On 12/09/2013 05:14 PM, Eric Moffatt wrote:
What happened to the idea of scraping github for data ? This seemed like a great idea to me when it first came up and may provide further insight into our users 'real' preferences.
Do you, as a user, commit your preferences to GitHub?
I may be wrong, but I don't believe GitHub repositories can give real hints about IDE configuration.
why don't you simply go and look?
For example, at github the option 'org.eclipse.jdt.core.compiler.problem.autoboxing' is set 14,244 times to ignore [1] (today's default), and only 1,061 to warning [2] (the proposal) and 230 to error [3].
That's much better data than the survey, because it is more detailed, is based on real usage, and we have seven times as much 'votes'.
And surprisingly (or not) the result is very different.
[1] - https://github.com/search?q=org.eclipse.jdt.core.compiler.problem.autoboxing%3Dignore&type=Code&ref=searchresults
[2] - https://github.com/search?q=org.eclipse.jdt.core.compiler.problem.autoboxing%3Dwarning&type=Code&ref=searchresults
[3] - https://github.com/search?q=org.eclipse.jdt.core.compiler.problem.autoboxing%3Derror&type=Code&ref=searchresults
_______________________________________________
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
|
