Stability of Squeak

Mark Guzdial guzdial at cc.gatech.edu
Tue Aug 14 13:21:37 UTC 2001


>On Tuesday, August 14, 2001, at 07:31 AM, robin wrote:
>
>>In my commercial work, I'm used to the idea that even code which has been
>>thoroughly tested, with a seemingly comprehensive set of test cases by the
>>developers (exaustive/exhausting testing), will often fall apart under
>>seemingly obvious condition when real users first use it.  With squeak, it
>>seems that the idea is that everyone is developer - the only people who are
>>'just users' are newbies.
>For me, at least, Squeak has been solid as a rock -- among the most 
>stable development platforms on which I have worked.  Frankly, the 
>instability you are reporting is foreign to me -- I simply never 
>experienced it, and neither have the people (professionals, newbies, 
>adults and kids) here who are working on Squeak.

Here at Georgia Tech, Squeak has a (relatively undeserved) reputation 
for being prone to crashes.  The cause, for the most part, is due to 
issues outside the software itself:
- Our students' filesystems are on UNIX boxes that get mounted via 
Samba onto the NT machines in our labs.  If students ignore our 
advice and run their images from their filesystems, network hiccups 
cause disconnects between the changes file and the image file. 
Students find the loss of their comments and local variable names 
frustrating.
- During the Spring semester, we had the students use the MPEG 
plugin, just as it was being released.  That was a strategic error on 
our part.  The memory bugs bit lots of students really hard and led 
to some of the most outraged students so-far.
- In a lab setting, we're having a hard time keeping up with the most 
recent VM.  As many may recall, some of the early NT VMs had problems 
with networking.  Andreas quickly fixed the problems, but it took a 
while for us to migrate the new VM to all the corners where the old 
VM had migrated (e.g., students making a copy of the old VM for 
themselves, but then unable to get networking to work).

While I agree with Andrew in that I relatively rarely hit VM crashes, 
I can understand Edmund's frustration.  The point, though, is that 
Andrew is right -- a careful description of the situation is 
necessary, because the things that can lead to a flake-out can be 
fairly distant from the base distribution.

Mark

--------------------------
Mark Guzdial : Georgia Tech : College of Computing : Atlanta, GA 30332-0280
Associate Professor - Learning Sciences & Technologies.
Collaborative Software Lab - http://coweb.cc.gatech.edu/csl/
(404) 894-5618 : Fax (404) 894-0673 : guzdial at cc.gatech.edu
http://www.cc.gatech.edu/gvu/people/Faculty/Mark.Guzdial.html




More information about the Squeak-dev mailing list