<html>
<head>
<meta content="text/html; charset=windows-1252"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Hi,<br>
<br>
This looks like a case where the automation fails. I'll look into
it.<br>
<br>
Berk<br>
<br>
On 06/30/2015 08:36 PM, Mark Abraham wrote:<br>
</div>
<blockquote
cite="mid:CAMNuMATB5dcBy-dmnriOdocAPJnyEEbQk3y7XWNxVBbQJV2eYw@mail.gmail.com"
type="cite">
<meta http-equiv="Content-Type" content="text/html;
charset=windows-1252">
<div dir="ltr">Hi,
<div><br>
</div>
<div>An unexpected issue came up while running my automated
tests of the RC tarball. Some newly-merged fatal errors that
try to stop people using mdrun inefficiently sometimes cause
the regressiontests to fail because they are defaulting to
using too many threads. See <a moz-do-not-send="true"
href="https://pastee.org/jgyae">https://pastee.org/jgyae</a></div>
<div><br>
</div>
<div>We do have machinery in regressiontests to automatically
re-try the test with fewer threads/cores/whatever, but only in
response to a set of known mdrun error outputs, so we could
add this particular error to that set.</div>
<div><br>
</div>
<div>But in this case, the note on line 43 suggests we should
improve the code leading to the fatal error so that we aren't
subsequently double-crossing ourselves. What do you think,
Berk?</div>
<div><br>
</div>
<div>(This didn't come up in per-commit testing because we limit
those to small numbers of cores because there's normally
multiple testing instances running simultaneously on the same
machine. We can probably do a better job of this after 5.1,
with a bunch of new hardware available.)</div>
<div><br>
</div>
<div>Mark</div>
</div>
<br>
<div class="gmail_quote">
<div dir="ltr">On Tue, Jun 30, 2015 at 4:28 PM Mark Abraham <<a
moz-do-not-send="true"
href="mailto:mark.j.abraham@gmail.com">mark.j.abraham@gmail.com</a>>
wrote:<br>
</div>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div dir="ltr">Hi,
<div><br>
</div>
<div>I'll ship release candidate for 5.1 today, once I've
sorted out release notes and put stuff on servers, etc.</div>
<div><br>
</div>
<div>The last possible behaviour changing patch for 5.1 is <a
moz-do-not-send="true"
href="https://gerrit.gromacs.org/4753" target="_blank">https://gerrit.gromacs.org/4753</a>,
if someone's inspired to make it work well. <a
moz-do-not-send="true"
href="https://gerrit.gromacs.org/4272" target="_blank">https://gerrit.gromacs.org/4272</a>
is too late, even if someone's inspired to split off part
of it.</div>
<div><br>
</div>
<div>release-5-1 branch is active on git, gerrit and
jenkins. Any remaining bug-fix patches on master branch in
gerrit may be cherry-picked to release-5-1 branch (on
gerrit or locally, as needed).</div>
<div><br>
</div>
<div>Business will proceed as usual in release-5-1 - such
branches are open only for all kinds of bug fixes,
addition of docs, addition of tests, and in all cases
aimed at stability and correctness.</div>
<div><br>
</div>
<div>master branch is open for normal business, but people
will probably be a bit quiet there for a while ;-)</div>
<div><br>
</div>
<div>We are likely to issue a new 4.6.x some time soon, as
we've had a number of mdrun issues fixed. (There's one
known bug outstanding to fix, hopefully Michael has time
real soon now.) Then release-4-6 branch will be
permanently closed. We cannot maintain three active
release branches and a development branch.</div>
<div><br>
</div>
<div>We'll do a 5.0.x soon also, then change the policy for
release-5-0 to "mdrun correctness bug fixes only" mode.</div>
<div><br>
</div>
<div>Thanks to all for the hard work!</div>
</div>
<div dir="ltr">
<div><br>
</div>
<div>Mark</div>
</div>
</blockquote>
</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
</blockquote>
<br>
</body>
</html>