[12839] in athena10

home help back first fref pref prev next nref lref last post

Re: A request for packages to be added to debathena-thirdparty

daemon@ATHENA.MIT.EDU (Myron Freeman)
Tue Sep 9 09:35:34 2014

Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Mac OS X Mail 6.6 \(1510\))
From: Myron Freeman <fletch1@eecs.mit.edu>
In-Reply-To: <7837905E-D1DC-4BC4-97E2-58991F27196C@mit.edu>
Date: Tue, 9 Sep 2014 09:35:22 -0400
Cc: Debathena Trac <debathena@mit.edu>
Message-Id: <B3A1EEC9-AF26-4DC2-B5A6-42042F70F14A@eecs.mit.edu>
To: Jonathan D Reed <jdreed@mit.edu>
Content-Transfer-Encoding: 8bit

Thanks.

-Fletch

On Sep 8, 2014, at 5:54 PM, Jonathan D Reed <jdreed@mit.edu> wrote:

> This change should go to -development (“alpha”) tonight, -proposed (“beta”) later this week, and if all goes well, be in production by Monday.
> 
> -Jon
> 
> On Sep 3, 2014, at 4:57 PM, Myron Freeman <fletch1@eecs.mit.edu> wrote:
> 
>> This is going to be so painful :-)
>> 
>> Well I can start with HSPICE since there's a class using it for the first problem set. It is missing the i386 version of libnspr4.so
>> 
>> -Fletch
>> 
>> On Aug 29, 2014, at 4:29 PM, Jonathan D Reed <jdreed@mit.edu> wrote:
>> 
>>> OK, let us know when you have a list of packages, and we can do it.  Our turnaround time for cluster machines is about 5 business days, assuming nothing goes wrong.  The dialups are on a less frequent schedule, but an update can generally be done in about a 2 week turnound time (2 weeks from when it hits production on the cluster machines)
>>> 
>>> -Jon
>>> 
>>> On Aug 29, 2014, at 4:24 PM, Myron Freeman <fletch1@eecs.mit.edu> wrote:
>>> 
>>>> 
>>>> On Aug 29, 2014, at 1:58 PM, Jonathan D Reed <jdreed@mit.edu> wrote:
>>>>> 
>>>>>> ia32-libs - At some point this was added to Precise but it looks like only a subset is available under Trusty
>>>>> 
>>>>> This package doesn’t exist anymore.  We have created debathena-athena-libraries-ia32 to replace it, and will happily add any :i386 versions of the libraries there, but we really wanted to avoid “just copy what ia32-libs looked like in precise”.  Is it easy to get a list of what specific libraries it wants?  “ldd” on the binary should be useful.
>>>>> 
>>>> 
>>>> I (and the TAs) will have to weave our way through a few binaries so this could take a little while.
>>>> 
>>>>>> libelf1:i386 - This exists under debathena on Trusty but a lot of the CAD tools look for the 32-bit library under /usr/lib32
>>>>> 
>>>>> Can you expand on this?  They are dynamic libraries, so I’d hope that the software would just let ld.so do its job.  Are they setting LD_LIBRARY_PATH or otherwise doing something weird?  While we can put a symlink it, I’m reluctant to do so for a dynamic library without better understanding exactly what’s failing.
>>>>> 
>>>> 
>>>> I'll have to look into what exactly is happening with this one.
>>>> 
>>>>>> A lot of the CAD tools are also hardwired to look for awk at /bin/awk so I'm requesting that be a link to /etc/alternatives/awk
>>>>> 
>>>>> We can make /bin/awk a symlink to /usr/bin/awk (which in turn is a symlink to /etc/alternatives) on cluster machines, however if you have a contact with the vendor, we’d appreciate it if you could report the issue to them (if you haven’t already) and encourage them to just use ‘awk’ in $PATH, if it exists.  I realize there’s little hope of it getting fixed, but we are trying to stick to our policy of not putting in hacks/workarounds unless the bug has been reported.
>>>> 
>>>> Sure we'll report it to the vendors but I wouldn't hold my breath for a positive result :-)
>>>> 
>>>> -Fletch
>>>> 
>>> 
>> 
> 
> 



home help back first fref pref prev next nref lref last post