pexpect: to upgrade or not to upgrade?

4 views
Skip to first unread message

Franco Saliola

unread,
Mar 6, 2010, 9:57:29 PM3/6/10
to sage-...@googlegroups.com
I recently stumbled over a bug in the pexpect module shipped with
Sage. It turns out that specifying the full path to a command doesn't
work; you get an UnboundLocalError exception. I've created a ticket
with an example:

http://trac.sagemath.org/sage_trac/ticket/8471

The question is, do we patch pexpect, or do we upgrade to the newest
version (we're shipping version 2.0 while the latest version is 2.3)?
I imagine that there could be some good reasons for not upgrading, so
the real question here is whether there are any objections to
upgrading.

Franco

--

Robert Bradshaw

unread,
Mar 6, 2010, 10:40:59 PM3/6/10
to sage-...@googlegroups.com

The most likely case is that simply no one's had the time or need to
upgrade (though this would be the place to ask). I would recommend
trying it and seeing if it goes smoothly or there are issues. (It
would be good to try out the optional interfaces as well.)

- Robert

William Stein

unread,
Mar 7, 2010, 12:22:19 AM3/7/10
to sage-...@googlegroups.com
On Sat, Mar 6, 2010 at 7:40 PM, Robert Bradshaw
<robe...@math.washington.edu> wrote:
> On Mar 6, 2010, at 6:57 PM, Franco Saliola wrote:
>
>> I recently stumbled over a bug in the pexpect module shipped with
>> Sage. It turns out that specifying the full path to a command doesn't
>> work; you get an UnboundLocalError exception. I've created a ticket
>> with an example:
>>
>>   http://trac.sagemath.org/sage_trac/ticket/8471
>>
>> The question is, do we patch pexpect, or do we upgrade to the newest
>> version (we're shipping version 2.0 while the latest version is 2.3)?
>> I imagine that there could be some good reasons for not upgrading, so
>> the real question here is whether there are any objections to
>> upgrading.
>
> The most likely case is that simply no one's had the time or need to upgrade
> (though this would be the place to ask).

No, that's not the situation. IMHO every pexpect version > 2.0 I've
tried has seriously sucked performance wise, so far. They did a total
rewrite which made a mess of things... Anyway, the last time we had
this discussion was about a year ago, so it's worth trying again
(maybe pexpect improved).

William

Robert Bradshaw

unread,
Mar 7, 2010, 12:42:29 AM3/7/10
to sage-...@googlegroups.com


It may be worth adding this blurb to the pexpect's SPGK.txt.

- Robert

Reply all
Reply to author
Forward
0 new messages