apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jeff Trawick <traw...@gmail.com>
Subject Re: svn commit: r1541143 - /apr/apr/branches/1.5.x/STATUS
Date Sun, 17 Nov 2013 18:53:23 GMT
On Tue, Nov 12, 2013 at 11:57 AM, <trawick@apache.org> wrote:

> Author: trawick
> Date: Tue Nov 12 16:57:58 2013
> New Revision: 1541143
> URL: http://svn.apache.org/r1541143
> Log:
> no showstoppers?
> Modified:
>     apr/apr/branches/1.5.x/STATUS
> Modified: apr/apr/branches/1.5.x/STATUS
> URL:
> http://svn.apache.org/viewvc/apr/apr/branches/1.5.x/STATUS?rev=1541143&r1=1541142&r2=1541143&view=diff
> ==============================================================================
> --- apr/apr/branches/1.5.x/STATUS [utf-8] (original)
> +++ apr/apr/branches/1.5.x/STATUS [utf-8] Tue Nov 12 16:57:58 2013
> @@ -82,12 +82,6 @@ Bundled with httpd:
> -    * Verify that new features are:
> -      + documented
> -      + implemented across major platforms (if applicable)
> -      + have at least minimal regression tests in place

Yeah, skip list rant :(

Adding this "showstopper" was probably questionable according to
groupthink, but removing it was terrible judgement according to
Jeffthink...  This perhaps unobvious reminder of the practice of
documentation and regression tests didn't gain any visible interest in
resolving the omissions, and the 1.5.0 cycle was going on and on, and I
didn't choose to prioritize resolution of this over other things I might
spend my time on or other issues I might feel the need as RM to nag about,
so I failed.

The seemingly uninteresting acts of writing basic API documentation and
testcases are important opportunities to think about appropriateness of
interfaces and perhaps even notice questions in the code about the
interfaces.  Going forward, I don't think there is any reason for new
features in trunk to remain there any length of time without at least a
rough sketch of documentation and example use in the form of testcases, and
certainly new API features should not be merged to a stable branch without
it.  I don't mean to pretend that every last interface is technically
capable of being tested in the test suite or even that it is appropriate to
require anything testable to be tested, but instead to point out that these
widely accepted goals that are largely met by the existing codebase should
not get a pass on any new features, that we shouldn't be in a position of
providing new features which are barely consumable.

> -
> -    * Makefiles for MSVC

Born in Roswell... married an alien...

View raw message