3.9a-7006: ":=" replaced by "_"

Hi,

in the latest alpha, ":=" are automatically replaced by "_" when saving code in a System Browser.
Is that the new default behaviour ?

On 08.03.2006, at 09:10, Nicolas Melin wrote:

>
> Hi,
>
> in the latest alpha, ":=" are automatically replaced by "_" when  
> saving code in a System Browser.
> Is that the new default behaviour ?

Not in my image. Do you have pretty printing turned on? Ar you using  
Shout?

Pretty printing for now does _ , it should be :=, though (will fix  
that), as that's the new
standard.

Before 3.9 goes beta, we will have relaced all the _ by :=, and  
everything will be fine.

       Marcus

Oh sorry, I'm actually using Shout... this does not actually happen in the base image.

On Wed, 8 Mar 2006 09:25:12 +0100, Marcus Denker <[hidden email]> wrote:

Marcus Denker wrote:

>
> Before 3.9 goes beta, we will have relaced all the _ by :=, and
> everything will be fine.
>
>       Marcus
>
What happend with '<-' ? I kind of liked it more than ':=' or '_'.

On 08.03.2006, at 09:38, Dan Corneanu wrote:

There is no backarrow on my keyboard, and there is no backarrow
in the ASCII font. Especially, it is *extremely* strange to just co-opt
the underscore. What if I want to use an underscore? What
about copy and paste? Ever tried to read an email containing Squeak
code?

So the re-defined _ was clearly a bug, and thus we fixed it.

Now _ looks very strange for assignment, so it's better to use :=.

Another reason: Most changesets that I integrated came with :=
over the last years, as people use Shout which silently replaces
_ with := by default, or use Squeak's pretty printer which printed
:= by default, too.

Now all this does not mean that I think that ASCII is the best
way to display code. But if we want to move forward on that
regard, let's do it for real, not with this strange _ thing.

I for one would like to have a system that does not save text
at all but has a higher-level representation and a very powerful
pretty-printer.

        Marcus

> I for one would like to have a system that does not save text
> at all but has a higher-level representation and a very powerful
> pretty-printer.

Like a tree ?

Marcus Denker wrote: Yes, you're right.

Dan Corneanu wrote: actually, i too would like to see '<-' as an alternative to ':=', it's
non-ansi, but looks more 'smalltalkish'... some votes?

Hi Viktor, you <[hidden email]> wrote:

Just in case it was not mentioned before: I suggest a set of TWO compiler  
(parser) related preferences

        [] := is assignment
        [] <- is assignment

Then, if both boxes are not checked, _ is the (good old fashioned)  
assignment (and, of course, newbees have something to ask for ;)

/Klaus

>
>
>

Am 08.03.2006 um 11:46 schrieb Viktor Svub:

I liked the idea that was floating about that we always use ':='  
internally (and presumably externally, like files or clipboard, too),  
but we render it in the browser however the user likes, and we insert  
it on whatever keyboard binding the user chose. Best of all worlds :)

- Bert -

I'd rather keep the back arrow, too, even if I had to learn a new multi-key
combination to enter it.

As for source text versus something "higher level", I'll take source, though I
have no objection to a Unicode format, as opposed to straight ASCII.  Of
course changing the image to be so totally non-ASCII might be problematic...

On Wednesday 08 March 2006 05:46, Viktor Svub wrote:
--
To see a need and wait to be asked, is to already refuse.

Marcus Denker wrote:

Actually, it's the other way round... early versions of ASCII (1963) had
the left and up arrow symbols. Only later (1967) they morphed into
underscore and circumflex accent (see
http://en.wikipedia.org/wiki/ASCII). Looks like the early Smalltalk
implementors had an ASCII-1963 manual floating around, or they were
printing out source code on older printers with those glyphs...
My first "real" computer (a TRS-80) also had left arrow and up arrow
symbols in its character ROM (and no lowercase letters until I soldered
an additional memory chip onto the 7-bit screen memory...), so it seemed
very natural to me when I was introduced to Smalltalk a number of years
later.

Cheers,
Hans-Martin

/me turns to a tattered and faded page of a disintegrating DOS manual
and reads from the holy text.

 "The number of the left arrow shalt be 27."
                -- the Gospel according to DOS, page 437.

/me powers on 486

/me makes exactly 6 keystrokes, counting the boot menu.

A left pointing appeareth on my screen.

A smalltalk source file encoded with a char 27 will display correctly in
a DOS text editor. (hold ALT then type "27" on the neumeric keypad, then
release the ALT key)

Okay, okay, Dos does everything faster... Most people can barely type
their user name on linux in 6 keystrokes much less hit the enter...

Because linux sucks, it cannot display this charactor, instead it will
be interpreted as a control charactor and lead to all kinds of mayhem. =(
(all code pages are inferior to 437).


--
Don't let your schoolwork get in the way of your learning.

http://users.rcn.com/alangrimes/

Hi all!

Bert Freudenberg <[hidden email]> wrote:
> I liked the idea that was floating about that we always use ':='  
> internally (and presumably externally, like files or clipboard, too),  
> but we render it in the browser however the user likes, and we insert  
> it on whatever keyboard binding the user chose. Best of all worlds :)
>
> - Bert -

Which makes me recall my Namespaces code which also relied on
"rendering" source code instead of showing it verbatim.
I intend to make a presentation on it yet again down in Brussels btw. :)

Anyway, I agree with you Bert.

regards, Göran

Yes! Mi first machine was also a TRS-80, but a CoCo (Color Computer). It had
left arrow and up arrow glyphs too.

To have lowercase, I wrote a character generator that worked in graphics
mode. Instead of just 32x16 upper case, I could do 51x24, 64x24 and 85x24 in
upper and lower case. Oh, those good old times when you could understand the
whole computer...

Cheers,
Juan

----- Original Message -----
From: "Hans-Martin Mosner" <[hidden email]>
To: "The general-purpose Squeak developers list"
<[hidden email]>
Sent: Wednesday, March 08, 2006 6:51 PM
Subject: Re: 3.9a-7006: ":=" replaced by "_"

I thought there was a fix to make underscore "_" and
left-arrow assignment distinct.

Will replacing all left-arrow assignment with := cause
hundreds of methods to lose their original
timestamp/author information?  If so, is that worth
this cosmetic change?

Thanks,
  Chris

Marcus Denker <[hidden email]> wrote:
It looks fine. I'm using celeste.
>
> So the re-defined _ was clearly a bug, and thus we fixed it.

Hmm, I believe you broke it.

I fixed the squeak scanner so it only accepts the back arrow for
assignment, thus making it more like the original ST-80 scanner. This
simplifies the grammar a little, removes complexity and thus adds beauty
to the system as a whole.

Furthermore, the back arrow expresses the assignment much better than
"=" or ":=", the latter being easily mispronounced as "equals" instead
of "becomes".

I really love the back arrow -- and a three button mouse, even if it
does not comply to current expectations or hardware.

Regards,

Wolfgang Helbig

Lets just use 0x1b for back-arrow and demand everyone start using
Codepage 437 again. =))))




--
Don't let your schoolwork get in the way of your learning.

http://users.rcn.com/alangrimes/

Alan Grimes wrote:

>Lets just use 0x1b for back-arrow and demand everyone start using
>Codepage 437 again. =))))
>  
>
I strongly object :-)
CP 437 is much younger and therefore less mature than ASCII-1963.
And obviously it has been created by stealing and putting together parts
of other character sets (see http://en.wikipedia.org/wiki/Codepage_437),
whereas ASCII-1963 seems to be a design-by-committee thing, which is
inherently better ;-P

Cheers,
Hans-Martin

Am 09.03.2006 um 18:49 schrieb Chris Muller:

> I thought there was a fix to make underscore "_" and
> left-arrow assignment distinct.

It is, and the left-arrow glyph is still in the fonts. It's just not  
used anymore.

> Will replacing all left-arrow assignment with := cause
> hundreds of methods to lose their original
> timestamp/author information?  If so, is that worth
> this cosmetic change?

It would be pretty trivial to preserve the timestamps when patching  
the methods. But my FixUnderscores package currently does not.