Quantcast

issues transferring to github

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

issues transferring to github

Roman Ivanov
Hello All admins,


lets divide work on issues transfer ... all incomplete and old ideas need to be closed, transfer to github only meaningful issues that you understand and it is reasonable to fix, you can start conversation with author of issue - if no reply we close issue.

I am already started with Bugs 
Daniil already started with Patches

Ivan, please start with "Feature Requests".

I created few new Labels in Github issue system: incomplete and approved.
So any time somebody create issue on Github is has no label, if idea is not described properly or any other problem we mark it as incomplete, if issue is correct and we approve it we mark it as "approved".
All incomplete ideas that older then 1 year(or any period we decide) will be closed.

Thanks,
Roman Ivanov.

------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: issues transferring to github

Roman Ivanov
 Hello Admins,

Please consider to transfer new Check ideas (only) from Checkstyle to Sevntu.Checkstyle project.
After all that Checks that we done in sevntu.checkstyle project I do not believe that it is possible to finish Check after first implementation and without proper/long testing on big ugly sources.

Lets make new ideas(Checks) to get maturity in sandbox project - sevntu.checkstyle.
We already able to be used in most famous IDE and maven and ant, and instable state for sevntu.checkstyle project is OK in comparison to main project Checkstyle.

Does it make sense ?

thanks,
Roman Ivanov

>Hello All admins,
>
>source:  https://sourceforge.net/p/checkstyle/_list/tickets
>destination:  https://github.com/checkstyle/checkstyle/issues
>
>lets divide work on issues transfer ... all incomplete and old ideas need to be closed, transfer to github only meaningful issues that you understand and it is reasonable to fix, you can start conversation with author of issue - if no reply we close issue.
>
>I am already started with Bugs 
>Daniil already started with Patches
>
>Ivan, please start with "Feature Requests".
>
>I created few new Labels in Github issue system: incomplete and approved.
>So any time somebody create issue on Github is has no label, if idea is not described properly or any other problem we mark it as incomplete, if issue is correct and we approve it we mark it as "approved".
>All incomplete ideas that older then 1 year(or any period we decide) will be closed.
>
>Thanks,
>Roman Ivanov.
>------------------------------------------------------------------------------
>October Webinars: Code for Performance
>Free Intel webinars can help you accelerate application performance.
>Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
>the latest Intel processors and coprocessors. See abstracts and register >
>http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
>_______________________________________________
>Checkstyle-devel mailing list
>[hidden email]
>https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
>


Роман Иванов
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134791&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: issues transferring to github

Roman Ivanov
bump!
I do need you opinion/vote.

Here is most recent example of new feature/patch request - http://sourceforge.net/p/checkstyle/mailman/message/31501278/ .
Such request will never by applied in such form - because code is awful, no documentations, no tests, no configuration for maven site , no simple way to test it. That lead to no intensive from admin to spend a time on such cases .... .

So we are in vicious circle  - no body contribute from outside because Admins does not apply patches, admin does not apply patches because afraid of awful code support and testing, Checkstyle project is not developing because Admin have not focused on it full time and nobody contribute from outside.

As said before, to apply new Check or feature from outside Admin need to spend significant amount of time to be sure that Check is make sense and do work well on huge amount of sources - so Admin will never apply new Check even it is reasonable and useful.

In my experience, even I did 10 code reviews of Check, the same amount of testing against almost all sources I have - Check still have Exceptions during execution on non trivial code after release. The only factor helps to mature Check is a time and being alive in sandbox extension that always do code validation whenever you do development during your work, the more code was validated the better.

We need to give community way to make changes, test them-self, apply to their work process, test one time, let use it to others and then do contribution to Checkstyle.  

If you know other way to make Checkstyle manageable please let me know.


------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: issues transferring to github

Ivan Sopov
This is exact case is unavoidable - emailing attached diff to checkstyle-devel mailing list will always be an option.

>>Ivan, please start with "Feature Requests".
This thread is marked as "task" in may mail-client.

>>Please consider to transfer new Check ideas (only) from Checkstyle to Sevntu.Checkstyle project.
>>After all that Checks that we done in sevntu.checkstyle project I do not believe that it is possible to finish Check after first implementation and without proper/long testing on big ugly sources.

Testing new checks in sandbox is a valid Idea, but if check is intended to appear in future in the main checkstyle project maybe we should keep proper task in the main tracker?

Thanks, Ivan Sopov. 



2013/10/10 Roman Ivanov <[hidden email]>
bump!
I do need you opinion/vote.

Here is most recent example of new feature/patch request - http://sourceforge.net/p/checkstyle/mailman/message/31501278/ .
Such request will never by applied in such form - because code is awful, no documentations, no tests, no configuration for maven site , no simple way to test it. That lead to no intensive from admin to spend a time on such cases .... .

So we are in vicious circle  - no body contribute from outside because Admins does not apply patches, admin does not apply patches because afraid of awful code support and testing, Checkstyle project is not developing because Admin have not focused on it full time and nobody contribute from outside.

As said before, to apply new Check or feature from outside Admin need to spend significant amount of time to be sure that Check is make sense and do work well on huge amount of sources - so Admin will never apply new Check even it is reasonable and useful.

In my experience, even I did 10 code reviews of Check, the same amount of testing against almost all sources I have - Check still have Exceptions during execution on non trivial code after release. The only factor helps to mature Check is a time and being alive in sandbox extension that always do code validation whenever you do development during your work, the more code was validated the better.

We need to give community way to make changes, test them-self, apply to their work process, test one time, let use it to others and then do contribution to Checkstyle.  

If you know other way to make Checkstyle manageable please let me know.


------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel



------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: issues transferring to github

Roman Ivanov
In reply to this post by Roman Ivanov
Hi Ivan,

>> Testing new checks in sandbox is a valid Idea, but if check is intended to appear in future in the main checkstyle project maybe we should keep >> proper task in the main tracker?

Yes, we could put in main Checkstyle issue/reminder to port/copy particular idea to main project.  But I do believe that even author of new Check will remind us to move him out of unstable project(our sandbox).
Sanbox project could make each year graduation voting - where admins decide what Check is mature enough to go to main project.
As result I would rather skip additional issue registering in main project. It will be too much of them. It will not be managable. But if  all admin vote for duoble issue registering - I am ok.

So useful and required checks will not be lost.

Thanks,
Roman Ivanov
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: issues transferring to github

Roman Ivanov
Admins, please read this thread, I need you reply.

------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60135031&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: issues transferring to github

Roman Ivanov
Hi Daniil,

>>  I don't think that many people will remind us  by email or will try to discuss with us somehow else
If author of Check is not care about it any more - this Check should stay in sandbox till Admins find it useful and mature enough to be ported in main project.

>> Most likely, developers will just create issues in main repo
Unrelevant or emotional issues will be closed right away. Authors throw code to us that they do not want to support and fix so we need to defend your project.

>> So, we will always have trash/duplicate issues and we will always have to manage them.
So now we have 400 issues to resolve and reply in main project - trush and duplicates are unavoidable.
Trash amount mean that Admins have no time to read community feedback.
Duplicates amount mean that issue is in demand.
 
>> I don`t see any opportunity to vote on GutHub
I do not propose to have smth more spesial then just mail-list thread there one Admin offer list of Checks to join main project. Checks that got votes of all active administrators are nominated for porting. Community will use same mail-thread to share their deman is any (but I doubt that any body will do this except for us).

>> I think, we should have an easily detectable explanaton on wiki about where
sure

Please lets decide what to do with new feature requests. We will never have enough time to test all thoroughly.
Who will test https://github.com/checkstyle/checkstyle/pull/19 not from code reading but on real project with massive code ?


Thanks,
Roman ivanov.

=====================
Hello,

> I do believe that even author of new Check will remind us to move him out of unstable project
 I don't think that many people will remind us  by email or will try to discuss with us somehow else (after such experience as postponing issues and patches for months or even years when Checkstyle was at SourceForge).
Most likely, developers will just create issues in main repo (and/or the same issues in sandbox repo, that will be a pain),  including issues about porting some existing sandbox checks to main project. Public GitHub repos doesn`t allow to grant  creation of issues to repo commiters only. So, we will always have trash/duplicate issues and we will always have to manage them.

> Sanbox project could make each year graduation voting
I don`t see any opportunity to vote on GutHub issues/sources except for making additional trash comments like 'up!' on them or using external voting systems like Google moderator (which is not very convenient IMHO). I also think that we should not deceive users by opening voting process, cause after voting is done we can (and sometimes we really will) ignore voting results and continue using our own opinion.

In summary, I think, that we should always expect users to create 'please port xxxx Check' (and other) issues (and all issues that they want) in main repo. And, of course, we should create such issues for checks that tested by us (admins) on production code and approved by ourselves. We can use the special 'port' label for managing such issues.

I think, we should have an easily detectable explanaton on wiki about where, when, and how third-party developers should create issues for different cases, espesially for case when developer wants to force us to port some checks to main repo.

About more convenient issue management - I think it should be the separate discussion.

thanks
Daniil
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60135031&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: issues transferring to github

Roman Ivanov
In reply to this post by Roman Ivanov
Hi All Admins,

Looks like there are nobody here .... or all does not care at all, 20 days passed with no discussion and proposals .
So I apply to workflow my decision about sandbox project usage for new Checks,

Followig pull request will be the last for new Check for Checkstyle
https://github.com/checkstyle/checkstyle/pull/19

and for stabilization period I will apply only bugfixes and java 8 :) proposals.

thanks,
Roman Ivanov
------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk
_______________________________________________
Checkstyle-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/checkstyle-devel
Loading...