This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: bug-fix reviews [was: Bug 4930]


On Fri, 2007-09-14 at 13:16 -0700, Vara Prasad wrote:
> Jim Keniston wrote:
...
> >>>      
> >>>
> >>Can we follow/change our review process to that of kernel where patch is 
> >>posted inline in the mailing list for review instead of posting/placing 
> >>elsewhere? I am hoping having patches readily available in the mail 
> >>client improves our review process and avoids costly mistakes.
> >>
> >>bye,
> >>Vara Prasad
> >>    
> >>
> >
> >It's definitely appropriate to attach the patch to the bugzilla.  To get
> >a review, you either need to ensure that appropriate reviewers are on
> >the bugzilla's cc list, or explicitly request a review, as Wenji has
> >done.  I don't find it inconvenient to follow a couple of links to get
> >at the patch.
> >
> >I prefer to see review accomplished via the bugzilla mechanism, so that
> >the thread is captured in the bugzilla.  Of course, this typically
> >restricts visibility of the review thread to the bug owner and the
> >bugzilla's cc list.  Pro: Uninterested people aren't bothered with
> >extraneous emails.  Con: (a) A relevant reviewer may be excluded from
> >the cc list.  (b) If the review thread forks off a thread that warrants
> >wider participation, somebody has to take the trouble to move that
> >thread out of the bugzilla.
> >
> >Jim
...
> 
> Well, review comments thread can be accomplished via mailing list as 
> well hence i don't see any advantage of doing it in the bugzilla.

Again, a primary advantage is that the review is captured in the
bugzilla.

> There 
> are lot more people looking at the mailing list then just bugs and 
> excluding them from the review to me is a big disadvantage.

Your mileage may vary, but I prefer not to be cc-ed on the minutiae of
bug fixes that aren't related to what I'm working on.  (I admire people
who can continuously monitor all of (say) LKML AND get other useful work
done, but I'm not one of them.  Even Linus has been quoted as saying
that LKML is the wrong forum for bug reports.)

> Coming to 
> providing a link to bugzilla and asking for review still has the draw 
> back of traversing the link.

That's a skill that most software engineers have mastered. :-)

> I think following the common open source 
> methodology of reviewing the patches in the mailing list gives us better 
> chance of good reviews and we don't have to ask developers to learn a 
> new process.

Bugzilla is also an open source methodology.  It provides very
straightforward mechanisms for posting patches and responding to them.
It's not exactly a new process.

The IBM team does a fair amount of code review off the stap list (and
outside bugzilla, if the code isn't bug-fix-related).  If stap
developers find this alarming (AND are willing to do actual code
reviews), we can move more of this onto the list.

> 
> bye,
> Vara Prasad

Jim


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]