since i forgot to assign hw4.html, hw4.html will be to finish lab4.html
on your own, due next monday.  i WILL look at the directories for your work.

btw, wolf is running an older version of perl (5.0) which has a different
syntax for open and close, and which doesn't seem to permit system or
exec calls to generate output to the browser.  

if you want the output, you have to
pipe it to a file (/tmp/...) and then open the file.  k9.cs and hydra.cs
(the cs/siesta server, where my subman runs) are running a newer 
version of perl (5.6) which permits system calls to generate output
that the browser can understand.

so i recommend you do all your perl work on k9.

perl0 is fixed on my subman page.  i had "man perfunc" instead of 
"man perlfunc" which was causing it to omit that section of the manual.

some of you also noticed that "repeat 100 perl foo.pl < mydata" doesn't
work.  you have to put "perl foo.pl < mydata" in a file, bar, and say
"source bar", or "repeat 100 source bar".


-----------------------old stuff--------------------------------------------


for those of you who will be trying to get hw3 done in the crunch,
let me advise you to test on small images.  wolf is a dual 466 cyrix,
first generation, with 384mb of memory.  your cgi's can comfortably
process jpg's and gifs around 20k, but not much more.  if you find
that there is a process running which is taking up 300mb+, it probably
needs to be killed.  you can see this by running "top" to see what
jobs are running.  if the owner is nobody, you need to look at
the process id, in the first column, e.g., 23145, 
and say system("kill -9 23145") somewhere in a cgi program.  the
reason for this is that you are cs363, but the job belongs to "nobody",
which is the apache server.

k9 is faster, a 900 amd k7, but also has a 384mb limit.

the problem is that associative arrays are space-inefficient, and
compressed images that are about 20k-50k can be 10-20mb in memory.
so your process ends up getting swapped to virtual memory and it
"never" finishes.




indeed, we will have labs on monday running from 11-1 and 1-3.
i don't think we need to be really clever about making sure 
enough people arrive at 11, and enough people leave at 1.  but
what we'll say is that we won't overflow or take chairs, and
if there are no places to accomodate students, new arrivals will
have to wait.  obviously the early bird gets more time and more
attention.

sorry for the late notice -- i hope at least one person reads this email
before/during class so it can be announced during prof. butler's lecture.



cs363 students: two things.

1.  please note there is a an assignment this week, hw2.html.
	if you finish in the next couple of days, email me so i know.
	i'm giving people the green light to look at posted solutions
	this time, so if you finish earlier, i want to know that.

	this should make the assignment easier for everyone, and
	i am much more interested in people learning this stuff
	well (there is a lot to learn, yes?) than amassing evidence
	for giving B's.  

	by the way, you might start thinking about final projects.  it's
	way too early to be serious about staring, but it will take a long
	time to ok all of the individual projects.  i suppose you could
	just surprise us during exam week, but i think you might benefit
	from our feedback.

2.  we are contemplating a split of the monday lab, to 11-1 and 1-3.
	(it would still be possible to come 12-2, but we would discourage it.)
	we aren't sure whether cec has the resources to accomodate this,
	but we thought we'd ask the students first.

	please go to cs/~loui/363s04/times.html and tell us which time(s)
	you could come to if we did this.