Re: review process (was: C++ PATCH: Unify name lookup 2/n)

[Daniel Berlin <dberlin at dberlin dot org>]

On Monday, March 31, 2003, at 01:43  AM, Phil Edwards wrote:

> On Mon, Mar 31, 2003 at 01:03:22AM -0500, Daniel Berlin wrote:
> > If our problem is really in reviewing significant patches (ie i'm
> > assuming we aren't dropping 3 line patches on the floor, and that
> > people just approve these while reading through email regularly), we
> > could track those that aren't approved in a day through bugzilla.
>
> I was going to ask you about this... would patches then be also treated
> as bug reports?

Yes. For patches that are fixes for existing bugs, you just add the 
patch to the bug report already existing, obviously.  For new features, 
we have a severity labeled "enhancement", and thus, even patches for 
completely new features might be the fixes for these bugs. And of 
course, if a bug  doesn't exist, we should have a bug for the new 
feature created at the same time as the patch that adds it, that 
describes what exactly this enhancement is providing.  We usually just 
have these one line or paragraph overviews in emails to gcc-patches or 
at the top of the included ChangeLog, which isn't quite where they 
really belong.  After all, how many times have you gone through 
searching the archives to try to find a patch for a given feature that 
you know was sent in the past few months, but can't remember if it was 
approved, and can't find it in the ChangeLog (because you don't 
remember what files it affected necessarily)[1]?
As a random thought, this would also make it easier, as a side effect, 
to list what features a given release added when it comes time to make 
the changelog.
:)

Of course, this all requires some automation to be truly useful

>  Or would it be a seperate "interface", e.g., patchzilla?
>
>
> > I can even work up some code so that requests approved this way get an
> > email generated to gcc-patches stating that the patch was approved.
>
> That would be nice.
>
>
> Phil

[1] Maybe it's just me here, but in these cases, I also generally find 
the full-text mail archive search feature usually worthless, and resort 
to going to the month by month indexes, sorted by subject or author, to 
find the patch.  The ft search brings up all kinds of emails that 
contain the words (or worse, sometimes huge patches that happen to 
affect a line containing the words you are looking for), when you only 
want to look at patches whose first paragraph or two contain the words, 
and were sent by a certain person.  Trivial if our significant patches 
that aren't immediately approved are bugs, since you can search the 
comments without searching the attached patch files, and can just regex 
the bug submitter as well at the same time on the query form.  If you 
have a date range, or know it was changed (IE approved or submitted) in 
the past n days, that's easy too.


> --
> If ye love wealth greater than liberty, the tranquility of servitude 
> greater
> than the animating contest for freedom, go home and leave us in peace. 
>  We seek
> not your counsel, nor your arms.  Crouch down and lick the hand that 
> feeds you;
> and may posterity forget that ye were our countrymen.            - 
> Samuel Adams