Re: Silence does not imply consensus.

[Roland McGrath <roland at hack dot frob dot com>]

There are valid points on all sides here.  

I was remiss in letting the thread go on for so long without
responding at all.  Ideally, nobody should be such a bottleneck.
OTOH, I have a large backlog of other threads that I'm failing to get
to and I think my reasons for that are quite understandable.  (In
short, many things need my input because of subtleties I know better
than others do, but libc discussion and patch review are low on my
list of job priorities.)  I tend to respond quickly to short threads
of short messages where my reply can be short.  When I spend time
clearing up the backlog, sometimes it consists of reading a long
thread full of long messages and then not responding at all, because
it turns out I have nothing I feel I need to add any more.  When I
suspect I'm going to have to write long messages, I tend to put it off
in favor of quicker tasks.

Siddhesh and Carlos were remiss in following up my earlier reaction to
the feature proposal as they did.  I said that there was a
controversial part of the proposal (magic environment variable) and a
noncontroversial one (new API call), and suggested that Siddhesh send
a separate patch for the noncontroversial change first (it could get
review on technical details and testing and go in with relatively
little delay or high-level discussion) and then follow up with the
controversial change (it could get discussion on the controversy in
the context of the more mundane technical details having been mostly
settled by the review of the first change).  Some time later, Siddhesh
followed up by saying that the noncontroversial part was something he
didn't really care about, and so he'd dropped it in favor of doing
nothing but the controversial change.  Carlos then gave a purely
technical review and nobody discussed the controversy at all.

Perhaps I was not clear in the nature of my initial objection, though
I think that I was clear enough.  Siddhesh responded as if my
objection had been, "This is too complicated;" and so he said, "Here,
this a bit simpler."  What I'd actually meant to communicate was,
"This is somewhat complicated, but moreover a separable part of it is
something I object to strongly while another part is something I don't
object to at all; so I'm asking you to split it up, and taking the
opportunity not to even get into the full details of my objection
since that can be a more focused discussion after the split, and
because this request means I can put off something that will take more
time and effort on my part."

I am concerned that my initial objection was essentially ignored by
what ensued.  It is entirely possible that I might have failed to
notice the change having gone in, and might not have noticed it until
reading over the NEWS file when we were very close to the next
release.  Reverting and re-opening the whole issue at that point would
have been much more disruptive.  That said, none of this is the end of
the world; I certainly never suspected any ill intent on anyone's
part; and it is ultimately my responsibility to keep track of what
changes are going in if I want to object to any of them.


Thanks,
Roland