<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Chet Mehta wrote:
<blockquote
cite="midOF10A7CCBB.F0416DAF-ON8625720D.006D5A91-8625720D.00722C46@us.ibm.com"
type="cite"><br>
<font face="sans-serif" size="2">Tziporet,</font>
<br>
<br>
<font face="sans-serif" size="2">I understand that OFED1.1-rc7 was
released
without the requested change below despite the fact that the request
came
in before the deadline. As things stand, userspace is broken for ehca
in
OFED 1.1. Was the request rejected because ehca is considered a
technology
preview for this OFED release and if so are there well understood
(documented)
rules/limitations on how their submissions are handled?</font>
<br>
<br>
<font face="sans-serif" size="2">I'm asking to (a) learn from this
experience
(b) understand what we could/should have done differently (c) help
avoid
issues of this type in the future for us or other contributors. I would
appreciate hearing back from you. Thx!</font>
<br>
<font face="sans-serif" size="2"> </font>
<br>
<font face="sans-serif" size="2">:Chet.</font>
<br>
</blockquote>
Hi Chet,<br>
I regret that some of the latest patches could not be included in the
final release, but the main issue was that the patches came too late. <br>
Please note:<br>
1. ehca was added to OFED 1.1 in RC3, 31-Aug
(<a class="moz-txt-link-freetext" href="http://openib.org/pipermail/openfabrics-ewg/2006-August/001234.html">http://openib.org/pipermail/openfabrics-ewg/2006-August/001234.html</a>)<br>
2. Most user space issues raised by IBM were reported only on RC7,
which was supposed to be the final RC, with no more major changes.<br>
We nevertheless accepted patches that you submitted at that late phase
which affected only ehca. However, the patch in question (on the
configure file for user space) touches ALL components of the userspace
build, and was a very large patch as well. At that late date (2 days
before final release!), it is impossible to start testing a change of
this magnitude on ALL platforms to verify that the change was not
harmful.<br>
<br>
I agree we should learn from this experience: in the future, <i>please
</i>start your testing as early as possible so that problems of this
nature are caught early in the development cycle, and do not need to be
addressed at the last possible moment.<br>
<br>
I am sure that in the next release things will go more smoothly since
this was IBM's first participation in an OFED release.<br>
<br>
Tziporet<br>
<br>
</body>
</html>