BUG: 2 buglets and a Nino PDA

"G–ran" Hultgren gohu at rocketmail.com
Mon Feb 1 13:29:56 UTC 1999


Hi all!

A tiny buglet found:
"gives nil:"
(RWBinaryOrTextStream on: 'abc') ascii next
"but this gives $a:"
(RWBinaryOrTextStream on: 'abc' from: 0 to: 3) ascii next

Another buglet - if you remove ascii from above, the second version
will not work - isBinary does not get initialized since it relies on a
#reset to be sent to the stream, which does not occur when using
#on:from:to:

A silly fix to the first buglet would be to add (this little readLimit
fellow sure does getting set a lot!):
ReadWriteStream>>on: aCollection
	super on: aCollection.
	readLimit _ aCollection size

(well, my image did not crash after this... :-)

But I must say that the ReadWriteStream's hierarchy makes me a little
bit confused. And if you start looking at instvar refs on readLimit it
gets quite messy, or perhaps I am not quite getting it.

I have not looked into other Smalltalks Stream hierarchies, but I
assume that ReadWriteStream dates back a while - anyone like to
comment on the rationale behind the hierarchy?

Another little funny thing:

Got the 2.2 VM running with the mini image on a Philips Nino - very
cool! It just worked right from the start. Impressed. Now I just have
to figure out how to press that Alt-key... Have not found out how to
do that under PalmPC-WinCE. Does anyone know?

Since my Nino has 8Mb of RAM and a flash disk with 20Mb I will try
some fatter images to see what happens. Morphic maybe? Hehe....
slooooow, I suspect - but would be nice to see if it could work.

"123 factorial" timed in at 1665 ms instead of 31 ms on my
233MhzPentium laptop... 52 times slower. Ok. But still - the UI stuff
feels quite ok! It is a little bit slow but selecting text, typing
etc. works fine. Hmmm, 52 times? Sounds a lot.

Note: I had to browse the Time class and add the #millisecondsToRun:
method since the debugger jumped up in my face, but hey, it felt so
good to just hack in a method and then get it working ON A NINO. And
the classic UI feels quite appropriate for such a small screen - just
a little bit annoying with all the panes resizing all the time... :-)

While browsing around the Stream hierarchy I came upon a few identical
methods (bytecode identical also) in different classes, which made me
wonder if perhaps there is a need for a tool for "discovering" stuff
like that? It probably means that there is some redundancy going on
that could be fixed. Or perhaps such a tool already exists?

Well, thanks for listening and my regards to the Squeak team. The
coming Squeaky months (Dan Ingalls list of stuff to wait for) sound
truly interesting - as always!

regards, G–ran

PS. What is the "correct" procedure for bug reports and fix proposals?
Just drop it here in the mailing list with a nice header like
BUG:etc.etc? DS


===
G–ran Hultgren   | goran.hultgren at inspire.se | OO Consulting
Inspire          | Hem: +46 8 54022950       | http://www.inspire.se
Karlav”gen 18    | Mobil: +46 708 813936     | icq#:6136722
114 31 Stockholm | "All my opinions are MINE, MINE, MINE!!!"







_________________________________________________________
DO YOU YAHOO!?
Get your free @yahoo.com address at http://mail.yahoo.com





More information about the Squeak-dev mailing list