Empty emails from phabricator

Phabricator seems to like to send emails with zero content to the list
with relatively high frequency, like this:

From: Ikhlas Ajbar <iajbar@codeaurora.org>
Subject: Re: [PATCH] Remove unused function HasPICArg().
To: iajbar@codeaurora.org, bcahoon@codeaurora.org
Cc: llvm-commits@cs.uiuc.edu
Date: Tue, 19 May 2015 18:29:30 +0000 (6 hours, 21 minutes, 32 seconds ago)
Reply-To: reviews+D9775+public+5835e7bc078b3eb5@reviews.llvm.org

REPOSITORY
  rL LLVM

http://reviews.llvm.org/D9775

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/

_______________________________________________
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

or this:

From: John Kåre Alsaker <john.mailinglists@gmail.com>
Subject: Re: [PATCH] builtins: Add __probestack support functions for x86
To: john.mailinglists@gmail.com
Cc: llvm-commits@cs.uiuc.edu, howard.hinnant@gmail.com
Date: Tue, 19 May 2015 19:13:07 +0000 (5 hours, 39 minutes, 29 seconds ago)
Reply-To: reviews+D9858+public+2b5918b1dd4651e5@reviews.llvm.org

REPOSITORY
  rL LLVM

http://reviews.llvm.org/D9858

EMAIL PREFERENCES
  http://reviews.llvm.org/settings/panel/emailpreferences/

_______________________________________________
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

Can we fix this? I don't even understand what these are supposed to be.

They happen for things like closing a revision, adding dependencies or subscribers. But I agree that they are pretty useless because they don't actually mention the action that triggered the mail but just contain the accompanying comment which may be empty.

- Matthias

They happen for things like closing a revision, adding dependencies or subscribers. But I agree that they are pretty useless because they don't actually mention the action that triggered the mail but just contain the accompanying comment which may be empty.

Last time I brought this up, I was pointed at Phabricator's Phabricator as the place to report this bug. I gave up after looking for the right place to report it in that, and not finding the obvious spot.

Jon

They happen for things like closing a revision, adding dependencies or
subscribers. But I agree that they are pretty useless because they don't
actually mention the action that triggered the mail but just contain the
accompanying comment which may be empty.

Insofar that they identify that something has happened they are useful.
Their current state is IMO only moderately useful since it's not clear what
triggered it.

-- Sean Silva

They happen for things like closing a revision, adding dependencies
or subscribers. But I agree that they are pretty useless because
they don't actually mention the action that triggered the mail but
just contain the accompanying comment which may be empty.

Jonathan Roelofs <jonathan@codesourcery.com> writes:

Last time I brought this up, I was pointed at Phabricator's
Phabricator as the place to report this bug. I gave up after looking
for the right place to report it in that, and not finding the obvious
spot.

I guess they might have meant here:

  https://secure.phabricator.com/book/phabcontrib/article/bug_reports/

Of course, the first four bullet points on how to report a bug are
things you can only do if you administrate the phab install, and the
actual link to file a bug requires a login.

I'll go ahead and file one though - more and more people are using phab
for llvm, so the problems with it are coming up more and more often.

Sean Silva <chisophugis@gmail.com> writes:

Insofar that they identify that something has happened they are useful. Their
current state is IMO only moderately useful since it's not clear what
triggered it.

I'm not convinced these are useful - they do more harm than good. I can
skim an email sufficiently that I know whether or not to spend time on
it in a second or two. For these, I have to follow a link, let the page
load, scroll down to the list that catalogues the entire history of what
has happened, correlate some timestamps, and finally close the browser
window and return to my email. This takes at least an order of magnitude
longer, which is absurd for a list with the volume of llvm-commits.

Justin Bogner <mail@justinbogner.com> writes:

I'll go ahead and file one though - more and more people are using phab
for llvm, so the problems with it are coming up more and more often.

https://secure.phabricator.com/T8269

Justin Bogner <mail@justinbogner.com> writes:

Jonathan Roelofs <jonathan@codesourcery.com> writes:

Last time I brought this up, I was pointed at Phabricator's
Phabricator as the place to report this bug. I gave up after looking
for the right place to report it in that, and not finding the obvious
spot.

I'll go ahead and file one though - more and more people are using phab
for llvm, so the problems with it are coming up more and more often.

https://secure.phabricator.com/T8269

They closed this, claiming that we have custom code that causes this
behaviour. So... where do I file the bug?

IIRC, Manuel Klimek was maintaining Phabricator in his spare time. I've CC'd him.

Thanks for pulling me in - I indeed thought I had eliminated those messages (and I haven’t seen them on cfe-commits). I must have missed something that is happening more often in llvm land…