[Solved] A Tip for beginners

[Ten_Mile_Hike]

Just a reminder that when iterating over a string and inserting or deleting characters you have
to use "Downto" not "To" to get your expected results when processing multiple characters.

In this case below the expected result is the removal of all '4' characters from the string.
ie. 'AB4C44DE' becomes ABCDE'.

I leave it as an exercise for the newbie to understand why this is so.

P.S. Yes; I know that there are a myriad of ways to attack this problem. This example is for newbies

Code: Pascal  [Select][+][-]

1. procedure TForm1.Button1Click(Sender: TObject);
2. var
3.   S: string;
4.   x: integer;
5. begin
6.  
7.   S := '412345446444474';
8.   for x := 1 to length(S) do if s[x] = '4' then Delete(S, x, 1);
9.   Memo1.Lines.add(S); //Result '123546447' is incorrect
10.  
11.   S := '412345446444474';
12.   for x := Length(S) downto 1 do if s[x] = '4' then Delete(S, x, 1);
13.   Memo1.Lines.add(S); //Result '123567' is correct
14. end;                

[tetrastes]

It's better to remind beginners to turn on checks (at least -Ciro) and use debugger while developing, because the line

Code: Pascal  [Select][+][-]

1. for x := 1 to length(S) do if s[x] = '4' then Delete(S, x, 1);

raises Range check error.

[Thaddy]

@tetrastes
You are not a beginner. Downto is better. Do you really want me to give you the answer? 

[tetrastes]

Of course downto is better, as to is simply buggy. And the answer is obvious after seeing range check error, particularly for a beginner.
Are you against using checks and debugger?

[Thaddy]

No, I am against re-allocations...

[cdbc]

Hi
@tetrastes:

downto is better, as to is simply buggy.

NO 'to' is not buggy!!!
The reason you can't use 'to' in that situation, is because you're deleting char(s) as you go, thereby changing the count of items, which in turn screws up your loop!!!
Regards Benny

[tetrastes]

@cdbc
If we can say "downto is better", why we can't say "to is buggy"? 
I think it was obvious, what these meant. Why deprive newbies of finding the answer themselves?

[dbannon]

If we can say "downto is better", why we can't say "to is buggy"? 

• Because its not buggy. It does exactly what it is designed and documented to do.
• Because labelling something as basic as that as buggy is bad for FPC image.

(I am not suggesting we should not call out real bugs for that reason !)

Davo

[cdbc]

Thanks Davo, exactly my thoughts 
Regards Benny

[tetrastes]

It's strange to me that someone understand this phrase verbatim. I think its meaning was clear in the context of discussion.
Of course I don't think that "to is buggy" by itself. And that "downto is better" always and everywhere. 

[Thaddy]

@ tetrastes

You still don't get it, do you?
Removing chars with downto does not re-allocate memory for the whole string.
It just performs a move on the string tail. There is no need for reallocmem. It shrinks.
Removing chars with to, from zero,  DOES re-allocate memory for the whole string every time and is therefor considerably slower, calls reallocmem/getmem, move and freemem on every occasion.
So much for the beginners!
There, I gave it away.
Basics, basics, basics.

It was a genuinely good question for beginners and then an old hand falls in the same trap and forces me to reveal the answer.

Shame on you 

(at least the beginners now have the explanation)

Otherwise it is my opinion that to is not buggy. (who invented that novelty?)
Actually that is fact, not opinion, but I was watching a documentary about Carthage, "Ceterum censeo Carthaginem esse delendam,"  .

[MarkMLl]

It's strange to me that someone understand this phrase verbatim. I think its meaning was clear in the context of discussion.
Of course I don't think that "to is buggy" by itself. And that "downto is better" always and everywhere. 

In that case choose your words better: "to is /inappropriate/ in this case" is the correct way of putting it, particularly in something which is labelled as being for beginners.

However I'd also point out that since we're discussing strings here, with no indication of the encoding, it's necessary to take into account that if UTF-8 characters are involved the result might still be unexpected.

MarkMLl

[Thaddy]

Oh, Mark,
He has no clue anyway, why bother to overload his brain? :'(

[LV]

This is excellent advice. But this

Code: Pascal  [Select][+][-]

1.   S := '412345446444474';
2.   for x := 1 to length(S) do if s[x] = '4' then Delete(S, x, 1);
3.   Memo1.Lines.add(S); //Result '123546447' is incorrect
4.  

is usually called a beginner programmer's mistake, not an FPC bug. Isn't it?

[tetrastes]

@Thaddy

Though you like to insult people, I'll still answer you.

1. I didn't give any answer on OP's question, so why "shame on me"? 

2. Your answer explaines nothing, though it contains much bragging, as usual. The code with "to" doesn't work not because of reallocations and other stuff (which is problem of inefficiency only). If there was implementation without reallocation, it still wouldn't work, because it is mathematically incorrect. And I won't write more, let beginners think themselves (though @cdbc gave the clue).

But to illustrate this:

Code: Pascal  [Select][+][-]

1.   S := '412345446444474';
2.   x := 1;
3.   repeat
4.       if s[x] = '4' then Delete(S, x, 1)
5.       else Inc(x);
6.   until x = length(S)+1;

This code works despite all reallocations, etc.