Re: ChangeLog entries (was: C++ codegen bug for switch statement(s))

Re: ChangeLog entries (was: C++ codegen bug for switch statement(s))

[Mike Stump]

> Date: Thu, 14 Sep 2000 00:54:36 +0200 (CEST)
> From: Gerald Pfeifer <pfeifer@dbai.tuwien.ac.at>

> On Wed, 13 Sep 2000, Mark Mitchell wrote:
> > I don't generally make testsuite ChangeLog entries.  Is that something
> > we're supposed to do?

> Well, I don't know! ;-)  But once I do, I'll update the web page,
> promised.

I don't see enough value in it when measured against the cost.  In a
well done testcase, there are very few edits required post writing it,
there usually aren't interactions with other files, the comment is
usually something like, {Added.,Removed.,Fixed.} lending little
additional information.  ChangeLogs predate things like public
anonymous cvs repositories.  One might say that the ChangeLog
mostly replicates the cvs log.

I'd say, let's mandate changelog entries for any framework fixes
(lib/*, g++.other/*.exp), let's skip entries that would just say,
added, and leave the remainder up to the person doing the work.


Now, if enough other people do see a value in doing it, we can do it,
I don't think it is all that much extra work.  Out of curiosity, has
anybody made use of what changelogs are there today?

Re: ChangeLog entries (was: C++ codegen bug for switchstatement(s))

[Mark Mitchell]

>>>>> "Mike" == Mike Stump <mrs@windriver.com> writes:

    Mike> I'd say, let's mandate changelog entries for any framework
    Mike> fixes (lib/*, g++.other/*.exp), let's skip entries that
    Mike> would just say, added, and leave the remainder up to the
    Mike> person doing the work.

This would be my preference -- it reflects the way I already do
it. :-)

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com

Re: ChangeLog entries (was: C++ codegen bug for switch statement(s))

[Michael Meissner]

On Wed, Sep 13, 2000 at 06:18:45PM -0700, Mark Mitchell wrote:
> >>>>> "Mike" == Mike Stump <mrs@windriver.com> writes:
> 
>     Mike> I'd say, let's mandate changelog entries for any framework
>     Mike> fixes (lib/*, g++.other/*.exp), let's skip entries that
>     Mike> would just say, added, and leave the remainder up to the
>     Mike> person doing the work.
> 
> This would be my preference -- it reflects the way I already do
> it. :-)

I actually would like a comment in each testsuite file as it is added stating
what the original problem was....

-- 
Michael Meissner, Red Hat, Inc.
PMB 198, 174 Littleton Road #3, Westford, Massachusetts 01886, USA
Work:	  meissner@redhat.com		phone: +1 978-486-9304
Non-work: meissner@spectacle-pond.org	fax:   +1 978-692-4482

Re: ChangeLog entries (was: C++ codegen bug for switchstatement(s))

[Mark Mitchell]

>>>>> "Michael" == Michael Meissner <meissner@cygnus.com> writes:

    Michael> I actually would like a comment in each testsuite file as
    Michael> it is added stating what the original problem was....

I think that is what the bug-tracking system should be used for.
Ideally, bugs would come in through GNATS, with (probably
unsimplified) test-cases.  Then, bug-fixers would fix the problems,
close the bugs, and put smaller test-cases in the regression test
suite.  Name the testsuite files after the bug numbers.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com