headers
Gilson Costa | 4 Aug 1999 17:00
Picon

Keyboard Trouble

I'm using an Intel computer (Pentium) with an international keyboard (us+ keytable).
The only key that doesn't work is the quote key, giving me a Yen (¥) sign. In my keyboard the same key is used  to
get double quotes.
Any ideas why this happens?
PS: ( I have tried to get the sources and compile them but all works the same way)

Sorry for my bad english!
Permalink | Reply | 2 comments |
headers
Stephen Pope | 13 Aug 1999 20:32
Picon

Exception Hierarchies

Hello all,

I just added a discussion of exceptions with the Squeak class libraries
to the discussion page (http://minnow.cc.gatech.edu/squeak.748) and
thought I might post it here for comment.

Exception Hierarchies

In order to prepare for a more sophisticated and better integrated
exception system in Squeak, we need to plan what exceptions will be
thrown by the system classes. Below are three exception hierarchies: the
current 4thEstate base system, my proposed extensions, and a more
complete example taken from the VisualWorks Non-Commercial system (with
some platform-specific ones removed) (not included in this email for
brevity -- see the swiki page).

Current "out of the box" TFEI Exceptions

  Exception ('messageText' 'initialContext' 'resignalException'
'handlerContext' 'tag' )
    BlockCannotReturn ('result' )
    Error ()
      Halt ()
      MessageNotUnderstood ('message' )
      MyResumableTestError ()
      MyTestError ()
      ZeroDivide ('dividend' )
    IllegalResumeAttempt ()
    Notification ()
      ExceptionAboutToReturn ()
(Continue reading)

Permalink | Reply | 0 comments |
headers
Shane Hayes | 17 Aug 1999 03:50

Re: New round of benchmarks

Did you want us to send value to the block?

----- Original Message ----- 
From: Stephen Pope <stp <at> create.ucsb.edu>
To: squeak <squeak <at> cs.uiuc.edu>
Sent: Monday, August 16, 1999 4:13 PM
Subject: New round of benchmarks

> Hello all,
> 
> I'd be interested in collecting a new set of benchmarks for Squeak 2.5
> on various platforms. Here's the previous data with updates for the two
> platforms I've tested it on. If people send me data, I'll merge it into
> a list and post it. Alternatively, someone could put it all on a swiki
> page...
> 
> 
> Simple Squeak Benchmarks
> 
> Send speed in sends/sec:
> 
> [ | r t |
> 3 timesRepeat:
> [t _ Time millisecondsToRun: [r _ 26 benchFib].
> Transcript show: (r // t * 1000) printString; space]]
> 
>  636000  637000  636000  -- G3/266 (tower)
>  636000  636000  654000  -- 8600/300
>  561000  561000  561000  -- 6500/300
>  561000  561000  547000  -- G3 (desktop)
(Continue reading)

Permalink | Reply | 1 comment |
headers
Paul McDonough | 17 Aug 1999 09:00
Picon

RE: Re: Exception Hierarchies

Rather than write an eloquent discussion of TFEI's opinions re: the recent
Exception discussion, I'll just point to Allen Wirfs-Brock's recent post. 
That pretty well goes for us too, and thanks Allen for saving us the
trouble of trying to figure out how to say it.  It's worth noting that two
specific areas for improvement Allen mentioned were highlighted as needing
further work in the readme that accompanied TFEI's original EHS release - a
pluggable default handler, and fitting in better with the debugger.  We do
not expect to get to either piece of work particularly soon, but both are
still "on the list".

The evolving Exception hierarchy looks like something we'd use, and we'll
continue chit-chatting with Stephen Travis Pope offline as the community
iterates toward an integration with the base Squeak image.  Thanks to
Stephen for spearheading this.

Finally, we have run the basic tests on Squeak 2.5, and the EHS seems to
work just fine with that release.

Paul McDonough,
speaking off-the-cuff for the TFEI Squeakers
Permalink | Reply | 0 comments |
headers
Alan Lovejoy | 16 Aug 1999 21:51

RE: Exception Hierarchies

> ** Original Sender: Florin Mateoc <Florin <at> integral.com>
> Just a short protest for having Halt and UserInterrupt as Errors. They are
> not errors and they should not be caught by a generic error handler.

I'll second this opinion.

I also did not notice anything corresponding to  VW's "NoHandlerSignal"
in the hierarchy.  It is essential in cases where you only want to handle an
exception if there is no other handler up the call chain.  The mechanism
is as follows: if the exception machinery finds no handler for an exception
after having searched the whole sender chain, it raises the NoHandler
exception (starting with the same initial activation frame where the original
exception was raised).  Only if there is no handler for the NoHandler
exception does the default ("emergency") exception handler get invoked
(and that's why the error message in this case is "Unhandled exception: ...").

The NoHandlerSignal idea was invented by Mike Malcolm, if I remember
correctly.

--Alan
Permalink | Reply | 0 comments |
headers
Bijan Parsia | 12 Aug 1999 14:48
Picon

RE: Eliminating assignments and variable syntax (accessors)

At 10:55 PM -0400 8/11/99, Lex Spoon wrote:

>Well, it's certainly open to discussion :)  I'd think it best to count the
>first indent as just part of method definition.  So to get a block inside a
>method, you'd have to add *another* indentation, like this:
>
>	myMethod
>		| x y z |
>			x := 1
>			y := 2
>			z := 3

Blargh! That's gross! Imagine the errors! You can't *tell*, without a
comment, whether this is a misindentation or a block. *Silent*, *SEMANTIC*
errors due to copy and paste are evil. Python doesn't have *that* problem.

I *think* that *mere* indentation doesn't generate a new block in python.
Indendation replaces the role of braces in C or begin...end in Pascal--you
have *keywords* indicating *what sort* of block you're dealing with
(neither has closures, even ersatz ones, so this problem doesn't crop up).
And, IIRC, you have a *colon* marking the start, e.g.,:

def foo(bar):
	if bar == 3:
		return bar
	else:
		return bar * 10

So, really, the opening delimiter of a block is something like *colon*
newline, rather than just new line. (I get bit by this all the time!)
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jeffrey J. Hallman | 27 Aug 1999 15:48

Can't get to Swikis

Both yesterday and today, I've been unable to
http://minnow.cc.gatech.edu/squeakDoc.1 or to
http://minnow.cc.gatech.edu/squeak.1 using Netscape 4.51 (thru a firewall at
work) or using Netscape 4.6 (direct internet connection at home), I get only
an error message that says: 

Message not understood: last

I've seen no mention of this on the mailing list.  Am I the only one
experiencing problems? 

Jeff
Permalink | Reply | 0 comments |
headers
Tim Olson | 25 Aug 1999 14:42

Re: Enhanced Newbie Debugger

> On Tue, 24 Aug 1999, Doug Way wrote:
> 
> >
> > I did notice that some messages didn't show their returned objects...
> > simple accessors such as Rectangle>>corner, and some binary selectors such
> > as + and *.  I'm not sure if this was intentional, but I don't think you'd
> > want to leave these out.

Ashley George Taylor replied:

> I'm aware of this, but not sure why this is happening. Any insight would
> be welcome. I thought it had to do with primitives.

Certain frequently-used selectors are compiled in-line to special byte
codes, instead of being compiled as message sends.  These special byte
codes check the operand classes and directly perform the operation if
the operands are of the correct class for the common case; if not, the
byte code causes a message send.

	-- tim
Permalink | Reply | 0 comments |
headers
Dean_Swan | 5 Aug 1999 19:00
Favicon

Re: Eliminating assignments and variable syntax (accessors)


From:  Dean Swan <at> MITEL on 08/05/99 01:00 PM

You know, with all the time and typing that everyone has been spending debating
this issue, the very same people probably could have implemented these
proposals.  Squeak includes the source.  Just subclass the compiler and
prototype the things.  The only way you'll ever really understand how the
various schemes function in practice is to implement them and use them.

Besides, excepting Object, Magnitude, Behavior, and a handful of other classes,
isn't most of the standard image just window dressing (pun intended) and/or
experimental gadgets anyway?  ;-)

                    -Dean Swan
                    dean_swan <at> mitel.com

P.S. I just design embedded DSP systems for a living, and Squeak/Smalltalk is
mostly a "hobby" for me, but a widely held point of view in the LISP community
is "use LISP to write the language that you want to solve your problem in".
Wouldn't this strategy work equally well for Squeak/Smalltalk?

My point is that, with Squeak/Smalltalk,  you can have your cake and eat it too,
so shut-up already and go write some code. :-)
Permalink | Reply | 0 comments |
headers
Tim Rowledge | 21 Aug 1999 03:45

Unix 2.5 sources - who wants'em?

Well, almost.

I've rebuilt the sources (interp.c, sq.h etc) and added clean stubs for the
new socket prims to Ian's sources for *nix. Since it appears that Ian is
out of town for now, is anyone collecting files on his behalf? Come to
that, would anyone like to actually implement the new socket prims for
*nix?

I have also added config related stuff for NetBSD (to suit my ChalTech ARM
NetBSD box) and NetWinder (an ARM Linux box) which ought to be folded in.

The only actual 'bugs' were:-
 - the primitive table has prim 182 pointing to a function that does not
seem to exist anywhere, even in the Mac sources. This has been reported as
a bug some time ago but it still crept into 2.5.
 - Ian had implemented ioLowResMSecs() in sqXWindow.c, but the latest sq.h
has a macro of the same name. Simply removing the function from sqXWindow.c
seems to work ok, but may not be his favoured approach.

So, is anyone collecting files? Does anyone else want to build the *nix 2.5
VM for their favourite machine? Are you back in touch with the world Ian?

tim

--

-- 
The purpose of computing is insight, not numbers.  - Hamming
Tim Rowledge:  rowledge <at> interval.com (w)  +1 (650) 842-6110 (w)
 tim <at> sumeru.stanford.edu (h)  <http://sumeru.stanford.edu/tim>
(Continue reading)

Permalink | Reply | 0 comments |

Gmane