Beginner question :) search and replace in String

Douglas Brebner <squeaklists@...> writes:

> Please bear in mind that natural languages have much more redundancy and
> less precision than programming languages.

You are right!

Personally I believe programming languages should strive to be as close to the
 the more stringent English of specification documents, and not to natural
 language with *all* its facets. The main line of argumentation in my previous
 mail was thus that the very argument of natural languages can be used in favor
 of my point as well :)

> In addition, programming languages have handy tools like autocompletion
> that natural languages don't, making abbreviations less useful :)

That argument keeps popping up (eg from Ramon just in parallel to this
 answer, and also before) but I am not sure if auto completion is the whole
 story. For example in Object-C you have to write

   @"hello " stringByAppendingString: @"worlds!"

 to concatenate strings. Whereas in Smalltalk we just write

   'hello ', 'worlds!'

 clearly both are as writable when you got auto completion, but isn't the
 second just more readable? In his "Elements of Style" William Strunk
 recommends to omit unnecessary word. With regard to source code, I'd say the
 lesson should be that more verbose is typically more readable but *not*
 always.

So given this observation that sometimes operators (or single letters, which
 for that sake the same) are *sometimes* more readable than natural language,
 we can go an revisit the set of operators in Smalltalk and check if there
 might be more candidates. And I'd be more than surprised if are were not some
 more candidates for abbreviation.

For example Pharo already got #<< for appending to a stream, so the same would
 make sense to append elements to a collecton, such that we can write

    list := List new << 'lorem' << 'ipsum' << 'dolor'.

et cetera.

--AA


_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Em 06/01/2010 11:18, Adrian Kuhn < [hidden email] > escreveu:

Two points on "more readable".  _Right now_ to Smalltalkers  that know
by heart that #, is a concatenation operator, it seems *less cluttered*
and thus it appears more readable.

Put forward some twenty to fifty years (or to do it right now ask a
girlfriend that doesn't know nor Objective C nor Smalltalk) to see
what of two expressions convene more meaning.

As everybody else knows :-P the string concatenation operator is '+',
right!?  So #, needs to be explained anyway. . .


This may lead to the same problem we already have with #, namely it is
slow and not recommended for repeated operations.

We have to strike a balance between the easiness for writing the code
with the implementation that will run it.

my 0.019999....

--
Cesar Rabak

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

On Wed, 6 Jan 2010, [hidden email] wrote:

You're wrong, it's '.'. :)

No, it's fast, because it doesn't create copies. Of course to know that
you also have to know that List is an alias for OrderedCollection...
It's just a C++ism that makes your code slower. :)


Levente

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Em 06/01/2010 17:30, Levente Uzonyi < [hidden email] > escreveu:

>  On Wed, 6 Jan 2010, [hidden email] wrote:
[snipped]
> > As everybody  else knows :-P the string  concatenation operator is
> > '+', right!?  So #, needs to be explained anyway. . .
>  You're wrong, it's '.'. :)

You know this polyglot approach to programming ends up with some
confusion :-)

[snipped]

> > This may lead  to the same problem we already  have with #, namely
> > it is slow and not recommended for repeated operations.
> >
>  No, it's fast, because it  doesn't create copies. Of course to know
> that   you  also  have   to  know   that  List   is  an   alias  for
> OrderedCollection...   It's  just  a  C++ism that  makes  your  code
> slower. :)
>
Levente,

In a ordinary (sort of, it's a dev image, Pharo 1.0 #10503, there is
no List class, and:

Collection>>, aCollection
        ^self copy addAll: aCollection; yourself

Which creates copies for each #, method send, right?  Or do I miss
something here?

--
Cesar Rabak

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

On Wed, 6 Jan 2010, [hidden email] wrote:

Yes you do, List is one of Adrian's ideas to Huffman code smalltalk.
Details here http://www.iam.unibe.ch/~akuhn/blog/2009/one-letter-method-names/


Levente

>
> --
> Cesar Rabak
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

Em 06/01/2010 22:06, Levente Uzonyi < [hidden email] > escreveu:
Ok Levente!

In this case after reading Adrian's blog post, I would say "List" is
not a good choice:  VW uses List for a different kind of object (a
descendant from OrderedCollection, in fact).

As a side note, IMNHO Object>>out (as Travis' idea) seems to me better
than Object>>p for the same purpose.

Regards,

--
Cesar Rabak

_______________________________________________
Pharo-project mailing list
[hidden email]
http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project