[7715] in testers
Re: My Athena session will not run.
daemon@ATHENA.MIT.EDU (William Cattey)
Thu Sep 18 18:42:06 2008
In-Reply-To: <200809182224.m8IMOFqd029673@outgoing.mit.edu>
Mime-Version: 1.0 (Apple Message framework v753.1)
Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
Message-Id: <53BF7262-EBEE-43F1-8556-A6C38FB901E0@mit.edu>
Cc: testers@mit.edu, Greg Hudson <ghudson@mit.edu>
Content-Transfer-Encoding: 7bit
From: William Cattey <wdc@MIT.EDU>
Date: Thu, 18 Sep 2008 18:41:18 -0400
To: athena10@mit.edu
There is one more issue:
I didn't know that by selecting "GNOME" from the list of sessions,
that I was changing how my GNOME session would work, and that it
would stop using the tcsh X session.
The greeter choices are:
Last session
1. Run Xclient script
2. GNOME
3. Secure Remote Connection
Failsafe GNOME
Failsafe Terminal
I guess choosing one of the two Failsafe options doesn't change the
default session.
But calling it "Last session" in the greeter caused me to figure I
had to choose GNOME
to undo having chosen Failsafe GNOME.
When I did that, I thought I was going back to the GNOME session I'd
been using, but apparently I created a new GNOME session that didn't
use tcshrc to do the work.
(Lucky break with my .path file.)
Should I have chosen "Run Xclient script"?
How will people who use GNOME as their UIMS know that the GNOME they
want is not GNOME, but is in fact Xclient??????
-wdc
On Sep 18, 2008, at 6:24 PM, ghudson@MIT.EDU wrote:
> Okay, mystery solved.
>
> 1. Bill had a .tcshrc file. In Athena 9.4, we have a local
> modification to tcsh which causes .tcshrc to be executed in addition
> to, rather than instead of, .cshrc, but that's gone in Athena 10. So
> .cshrc wasn't getting run. We should document this.
>
> 2. Bill has a .path file. The handling of .path files in the Athena
> 10 cshrc is buggy, and caused the "set: Variable name must begin with
> a letter" message. I've committed a fix.
>
> 3. xsession.tcsh explicitly sources .cshrc, which bypasses (1) and
> runs into (2), crashing the session.
>
> 4. At some point during testing, Bill chose the "gnome session", as
> opposed to the default. For non-default sessions, we don't execute
> xsession.tcsh, so the session did not crash out. It turns out picking
> a different session is sticky (saved in ~/.dmrc), so he was able to
> continue to bypass (2) for subsequent logins. (Not running the Athena
> xsession wrapper is fairly subtle; you don't get the gmotd, zephyr
> isn't started, your environment setup happens in each terminal instead
> of once globally, and the "logout" alias won't work, but none of those
> particularly jump out.)
