OpenEuphoria

1. problems with the scoping in eu4

• Posted by bill Jan 07, 2011
• 2192 views

1 Shadowing:

One cannot redefine variable names eg

integer f = 0 
... 
sequence f = {} 

is an error

likewise :

for i = 1 to 10 do 
  puts(1,sprintf("%d, ",2*i) 
end for 
... 
for i = 1 to 20 do .. <= error 

however this is different:

integer f = 2 
function ff(sequence f = "ok") 
  return f 
end function 
puts(1,sprintf("%s\n",ff())) => ok 
puts(1,sprintf("%d, ",f)     => 2 

2 while do problems:
Cannot initialise a counter inside a loop

integer r = 4 
while r > 0 do 
  integer i = 0 
  puts(1,sprintf("%d, ",i) 
  i += 1 
  r -= 1 
end while 

prints: 0, 0, 0, 0

while r > 0 with entry do 
  integer i = 0 
  puts(1,sprintf("%d, ",i) 
  r -= 1 
entry 
  i += 1  <= at runtime i has no value 
end while 

3 while with entry seems quite unnecessay given exit (using ada loop, end loop & exit when).

    sequence ret = {} 
    loop  
	   from = find_from (x, source, from) 
	   exit when from = 0 
       ret &= from  
       from += 1  
    end loop  

Exit can take a label.

Now the code executes in order, there is no jump into scope.

4 another reason not to use while with entry (readability).
It places the entry code below any inner loop

5 Switch has problems as well.

switch a 
case x then 
  integer i = 0 
  puts(1, sprintf("%d",i)) 
  i += 1 
case x+1 then 
  puts(1, sprintf("%d",i)) 
  i += 1## 
... 

This should compile but cannot work with a = (x+1)
Any declaration before case else allow a jump into scope.

6. Commentary:
ada has anonymous blocks syntax
eg an in-line swap of integers x, y

declare 
  t integer := x 
begin 
  x := y 
  y := t 
end 

A possible while syntax could be

while condition 
declare  
  type id [= value] 
do 
  .. 
end 

I am not saying this is how things should be done but these aspects of Euphoria make the new features difficult to impossible to use.

[matt: added eucode tags]

3. Re: problems with the scoping in eu4

• Posted by coconut Jan 07, 2011
• 2102 views

integer r=5 
while r do 
  integer i=0 
  printf(1,"%d, ",i) 
  i+=1 
  r-=1 
end while 

Bill raise a strong point here.
What's the point of beeing able to declare a variable inside a loop construct if its value is reset at each loop cycle?
It would be better to forbid variable declaration inside loop as it was in version 3.x
It is what I suggest concerning variables declarations: a rollback to the situation of version 3.x

And I very like the

exit when condition  

this could be added to euphoria 4.x

Jacques

5. Re: problems with the scoping in eu4

• Posted by jaygade Jan 07, 2011
• 2081 views

integer r=5 
while r do 
  integer i=0 
  printf(1,"%d, ",i) 
  i+=1 
  r-=1 
end while 

exit when condition  

Do this:

if condition then exit 
end if 

for i = 1 to 10 do
  puts(1,sprintf("%d, ",2*i)  
end for  
...  
for i = 1 to 20 do .. <= error 

I agree that this shouldn't be an error. It never was in the past, has it become an error now?

integer r = 4
while r > 0 do  
  integer i = 0  
  puts(1,sprintf("%d, ",i)  
  i += 1  
  r -= 1  
end while 

while r > 0 with entry do
  integer i = 0  
  puts(1,sprintf("%d, ",i)  
  r -= 1  
entry  
  i += 1  <= at runtime i has no value  
end while 

I see your point with regards to the first one. This should probably be coded as follows: (untested)

while r > 0 with entry do 
    integer i 
    puts(1, sprintf("%d, ",i) 
    r -= 1 
entry 
    if object(i) = 0 then i = 0 -- test and initialize 
    end if 
    i += 1 
end while

Although the test does kind of break the utility of "with entry". Maybe there's an even better way of coding that?

Heh - could you do this?

i += object(i)

Probably not.

sequence ret = {}
    loop   
	   from = find_from (x, source, from)  
	   exit when from = 0  
       ret &= from   
       from += 1   
    end loop  

Enough people wanted "with entry" so the devs added it. Enough people found it clearer than using a do loop. And entry works with do loops as well. To each their own.

7. Re: problems with the scoping in eu4

• Posted by Vinoba Jan 07, 2011
• 2075 views

Why not allow |if condition then i = i+7 |

Or the simple question is: Why do developers want to put more restrictions and formalitites instead of allowing programmers idiosyncracies?

9. Re: problems with the scoping in eu4

• Posted by jeremy (admin) Jan 07, 2011
• 2082 views

I'm lost in this thread, what exactly are you speaking of that was added that does not provide any benefit? Are you speaking of:

sequence names = { "John Doe", "Jim Doe", "Jack Doe", "Joe Doe" } 
 
for i = 1 to length(names) do 
    sequence name = names[i] 
    sequence firstName = get_first_name(name) 
    sequence lastName = get_last_name(name) 
 
    if equal(lastName, "Doe") then 
        printf(1, "CAUTION: Last name is really not known\n") 
    end if 
 
    printf(1, "Dear Mr. %s,\n\nThank you for your suggestion %s!", { lastName, firstName }) 
end for 

i.e. declaring variables inside of a loop and those variables not persisting? If so (I think that is what we are talking about), that feature brings huge benefit into clean function design. I really hated not being able to do it 3.1. Being able to is a big step ahead for Euphoria. It would be a terrible thing if this functionality were to be reverted.

If you want a variable to persist, then you declare it in the scope it should persist in. For example:

sequence names = { "John Doe", "Jim Doe", "Jack Doe", "Joe Doe" } 
sequence lastNameProcessed = "None" 
 
for i = 1 to length(names) do 
    sequence name = names[i] 
    sequence firstName = get_first_name(name) 
    sequence lastName = get_last_name(name) 
 
    if equal(lastName, "Doe") then 
        printf(1, "CAUTION: Last name is really not known\n") 
    end if 
 
    printf(1, "Dear Mr. %s,\n\nThank you for your suggestion %s!", { lastName, firstName }) 
 
    lastNameProcessed = named 
end for 
 
printf(1, "Sent %d messages, last name processed was %s\n", { length(names), lastNameProcessed }) 

This is clean design and I'd suggest that everyone begin programming by keeping the variable to the minimum scope necessary.

Jeremy

11. Re: problems with the scoping in eu4

• Posted by jaygade Jan 07, 2011
• 2042 views

sequence names = { "John Doe", "Jim Doe", "Jack Doe", "Joe Doe" } 
 
for i = 1 to length(names) do 
    sequence name = names[i] 
    sequence firstName = get_first_name(name) 
    sequence lastName = get_last_name(name) 
 
    if equal(lastName, "Doe") then 
        printf(1, "CAUTION: Last name is really not known\n") 
    end if 
 
    printf(1, "Dear Mr. %s,\n\nThank you for your suggestion %s!", { lastName, firstName }) 
end for 

sequence names = { "John Doe", "Jim Doe", "Jack Doe", "Joe Doe" } 
sequence lastNameProcessed = "None" 
 
for i = 1 to length(names) do 
    sequence name = names[i] 
    sequence firstName = get_first_name(name) 
    sequence lastName = get_last_name(name) 
 
    if equal(lastName, "Doe") then 
        printf(1, "CAUTION: Last name is really not known\n") 
    end if 
 
    printf(1, "Dear Mr. %s,\n\nThank you for your suggestion %s!", { lastName, firstName }) 
 
    lastNameProcessed = named 
end for 
 
printf(1, "Sent %d messages, last name processed was %s\n", { length(names), lastNameProcessed }) 

I don't think that the complaint is persistence outside of the loop. After all, that doesn't even make sense.

I think that the complaint is persistence from iteration to iteration. The variable is either unassigned, or it is reassigned some default value on every new iteration of the loop.

The simple example initially given was a good one, trying to use two counters in one loop.

13. Re: problems with the scoping in eu4

• Posted by mattlewis (admin) Jan 07, 2011
• 2009 views

In that case, it's simply misunderstanding how lexical scopes generally work. For example:

#include <stdio.h> 
 
int main(){ 
	int i = 5; 
	while( i ){ 
		int j = 0; 
		printf("--i: %d  ++j: %d\n", --i, ++j ); 
	} 
	return 0; 
} 

Here's what happens:

$ gcc -oloop loop.c && ./loop 
--i: 4  ++j: 1 
--i: 3  ++j: 1 
--i: 2  ++j: 1 
--i: 1  ++j: 1 
--i: 0  ++j: 1 

A variable goes out of scope when the end of its scope is reached. In this case, the scope is the body of the loop. The original request seemed to me to be for the equivalent of C's static (function-level) variables inside a particular scope.

Matt

15. Re: problems with the scoping in eu4

• Posted by mattlewis (admin) Jan 07, 2011
• 2000 views

It's similar, but different. exit jumps out of loops (for, while, until), but break jumps out of conditional blocks (if, switch).

It's somewhat subtle, but useful.

Matt

17. Re: problems with the scoping in eu4

• Posted by coconut Jan 07, 2011
• 2011 views

Yes! static variables inside function would be good, as Jeremy hammer it: a variable must have shortest scope as possible. And prensently in euphoria when one need a static variable in a function it must declare it outside the function. (but after rethinking it I wouldn't add this feature to loop structure).

I will add it as a feature request

Jacques

P.S. feature request #588 created.

19. Re: problems with the scoping in eu4

• Posted by jaygade Jan 07, 2011
• 2035 views

Edit: never mind. The subject changed.

Just what Euphoria needs, yet another storage class.

21. Re: problems with the scoping in eu4

• Posted by bill Jan 07, 2011
• 1983 views

To summarise, clarify and re-emphasise points:

I am not particularly concerned about syntax. I am concerned about meaning.

1 Inability to reuse for-loop variables:

The issue here is that using i,j,k for loop indices is common practice. It's universal mathematical use just as a,b,c are constants and x,y,z are variables./
The problem in Euphoria is that when the variable has gone out of scope the name and it's type hasn't. That says to me broken scope.
This is emphasised when we see that function parameters do shadowing correctly. The parameter can be of any type and it eclipses the outer declaration only while it is in scope.

2 While loops:

This isn't about static variables. It is about continued re-declaration.
The issue of reusing variables should imply that it is illegal to create variables in a loop as the name will not be released after use.

To say that the variable doesn't persist from loop to loop (Matt) is wrong - the variable should not go out of scope until the loop is exited. Therefore either the variable creates a re-declaration error as in 1 or the the declaration line has to be skipped on subsequent loops.

Initialisation to a constant should be possible if variables can be created in a loop.

One *could* enclose the declaration in an if-test but that is very defensive and is an avoidance of the issue

What should happen is that the the declaration is treated as part of the initialisation of the loop.

3. While with entry:

Starting at the bottom of a loop to avoid initialisation problems is bad design. Performing a sequence of statements out of order is *unstructured*. In the context of variable declaration it is a *jump into scope*.
It will break things.
And it is picked up by the compiler.

Starting at the bottom could mean starting 200+ lines away from the head of the loop (really bad stuff - but possible).

Nesting such loops makes for a kind of spaghetti. Stick a large switch in it and you have a complete mess.

Many may have voted for while with entry as a cute idea based on a single loop which could have been coded as a loop .. exit on condition ... end loop structure. I don't think they were voting for its use with variable declaration, nesting etc.

4. Switch:

No one seems to have commented on this.

Switches tend to be long as they enumerate the values the switch variable can take.

Declaring variables in a switch means they will be in scope over a large piece of terrain and multiple case labels. Jumping into scope will certainly happen. Fall through will allow other ways of coming into scope.

If variable declaration is outlawed in switch it would be no loss. However it raises an issue which is what to do if you want a local variable in a switch.

5 Blocks:

In order to limit the scope of variables it is necessary to be able to declare block scope. Currently that means: use a do once loop, or an if true then .. end if block or a procedure.

A block is conceptually very simple.
Example: in-line swap:
assuming two integers x, y
begin
integer t = x
x = y
y = t
end

If you like it is C's { .. } only {} means sequence.

If there were some expensive procedure to be done a block allows the space to be reclaimed immediately the block is complete.

Clearly blocks like this make more sense than declaring in loops or switches unless loop wide, switch wide scope was needed.

Having its own notation is cleaner than perverting loops or conditionals to this use.

Conclusion:

But first scope needs to be fixed.

23. Re: problems with the scoping in eu4

• Posted by bill Jan 07, 2011
• 1947 views

All I would really say is the scoping together with, the issue of name, type retention after the scope is exited make a situation where odd things happen.

I would not say that scope is broken because the issues are not that severe.

Not being able to declare a variable in a nested scope of the same name as a variable in outer scope is odd. Being able to declare a parameter of the same name as the outer variable (and for that to operate exactly as one would expect - temporarily shadowing the outer variable) makes things rather more odd.

None of the problems I have mentioned are show stoppers. I think they add complexity and particularity. Other than that well it is certainly programmable and rather elegant.

I guess the problem is elegant designs are often hard to improve.

I dislike while with entry and I think that variable declaration should be prohibited in while and switch, and that Eu should have a clean block scope notation.

25. Re: problems with the scoping in eu4

• Posted by bill Jan 07, 2011
• 1922 views

The problem I have is this: I write a lot of parsing code the structure of the code goes something like:

loop
get c
sym = symbol[c]
case c of
...
reading various type of data into variables etc
...
end case
...
end loop

Therefore I am not thinking of small loops with quirky behavior. Usually I don't use for loops because running off the end of a string is likely behaviour given bad data to parse. Using a while loop implies there is is a condition to test for - usually testing what the next character is.

Using an infinite loop is a good solution because the test(s) can be put where they are needed. Thus the motivation behind with entry do is always in the code.

Starting at the bottom of the loop is enticing if you only have while loops and no exit mechanism. Given these with entry do is pointless and potentially obscure.

27. Re: problems with the scoping in eu4

• Posted by mattlewis (admin) Jan 07, 2011
• 1947 views

I don't know where you got this idea. You can reuse for loop variables as much as you want.

Well, you may think that's how it should work, but I disagree.

[quote bill]

3. While with entry:

Starting at the bottom of a loop to avoid initialisation problems is bad design. Performing a sequence of statements out of order is *unstructured*. In the context of variable declaration it is a *jump into scope*.
It will break things.
And it is picked up by the compiler.
Starting at the bottom could mean starting 200+ lines away from the head of the loop (really bad stuff - but possible).

It's a choice of evils I suppose. Allow coders to avoid repeating themselves, or adhere to super strict structured programming paradigms. A 200+ line loop is bad stuff no matter what structured or unstructured programming is involved.

They may have, but I didn't. You either change the flow of the code, or move the condition to the bottom, which for me at least, makes the loop more difficult to read. Sorry, I really can't figure out what this has to do with the additional, smaller scopes.

It is possible that no one has commented because they could not figure out what you were talking about. Again, it's certainly not euphoria 4.0. For example:

-- bug.ex 
integer c = getc(0) 
switch c do 
	case 'a' then 
		integer x = 1 
		? x 
	case 'b' then 
		? x 
end switch 

$ eui bug 
bug.ex:8 
<0074>:: Errors resolving the following references: 
    bug.ex (8): x 
 
        ? x 
           ^ 

I recall talking about doing this, though we settled on using the blocks defined by existing euphoria flow control. The then/do...end blocks define our scoped blocks. I might be worth revisiting adding an explicit block. I don't recall why we decided against it.

While I admit that there are additional ways in which euphoria scopes could be enhanced or refined, I really don't see any existing bugs that need fixing. In fact, in several cases, euphoria already does what you said it should, and trying to do what you think it does will result in an error.

Matt

29. Re: problems with the scoping in eu4

• Posted by bill Jan 08, 2011
• 1947 views

About the switch:

I may have mis-read but I believe it is what the manual said.

I am interested in Euphoria because it is simple and fast enough for what I want.

I don't want to get into a big dispute about features and mis-features.

About delarations in loops I am clearly of the opinion it is a mis-feature. Perhaps I am in a minority of one thinking this. Also I think while with entry is a mis-feature again it is my opinion.

If I use Euphoria I will not use these features.

31. Re: problems with the scoping in eu4

• Posted by bill Jan 08, 2011
• 1827 views

integer cnt = 1 
for i = 1 to 10 do 
    cnt += 1 
    ? cnt 
end for 

for i = 1 to 10 do 
    integer ten_i = i * 10 
    ? ten_i 
end for 

1 while f > 0 with entry do 
2   integer sum = 0 
3   ... 
4 entry 
5   f = ... 
6   sum +=  
7 end while 

if TRUE then                    -- begin block 
  integer sum = 0 
  while f > 0 with entry do 
    ... 
  entry 
    f = ... 
    sum += f 
  end while 
end if                           -- end block 

if TRUE then 
  integer sum = 0 
  while TRUE do 
    f = ... 
    sum += f 
  exit when ... (or if f=0 then exit end if) 
    ... 
  end while 
end if       

if TRUE then                    -- begin block 
  integer sum = 0 
  while f > 0 with entry do 
    ... 
    if TRUE then                 
      integer x           -- declaration in the if-block not the while 
      ...                 -- will this pollute the while's namespace? 
    end if 
    ... 
  entry 
    f = ... 
    sum += f 
  end while 
end if                           -- end block 

declare 
  [declarations] 
begin 
  [run this code] 
end 

declare  
  integer sum = 0 
begin 
  loop 
    f = ... 
    exit when f = 0 
    sum += f 
    declare                 
      integer x  
    begin	   
     ...                      
    end 
    ... 
  end loop 
end 

You are wrong. If a loop is nothing but a goto then there is nothing sepecial about a loop. It is true a loop could be:

:loop 
  if f = 0 goto end 
  ... 
  goto loop 
  ... 
:end 

And you can merrily jump around through your code. You had better have global scope and a good debugger as you need both.

It is like that in sed and other obscure languages - (obscure as in difficult to understand).

However Euphoria loops have scope. There your argument falls down.

Declaring variables in while loops is not helpful.
With entry is an unneeded kludge presumably brought into existence by the lack of an exit statement.
With/without fallthru is undesirable - switch should have.

Surely

loop 
  [run this code] 
  [check for exit] 
  [run more code] 
end loop 

is less cumbersome, difficult to undestand and problematic than

goto x 
loop  
  [check for exit] 
  [run more code] -- declare and use loop declared variables here 
:x 
  [run this code] -- no loop variables here 
end loop 

Bill

33. Re: problems with the scoping in eu4

• Posted by jaygade Jan 08, 2011
• 1804 views

Umm, how is it unhelpful to be able to declare variables in while loops? I mean, it's not a feature that I would necessarily use and I think that it would be enhanced by being a true masking scope, but even without that I can see the utility of it.

With regards to "with entry" and the lack of an exit statement, umm, an exit statement exists. It is very flexible. You can even follow it with a certain number representing the loop depth or a specific label representing a specific loop. I'm not sure if any other language is so flexible.

With regards to switch statements and with or without fallthru, well, your statement was cut off. Can you explain better? People in the community could not decide whether a C-style switch (with fallthru) or a BASIC-style switch (without fallthru) so both were added, with the BASIC-style being the default.

35. Re: problems with the scoping in eu4

• Posted by jaygade Jan 08, 2011
• 1763 views

Just saying, not trying to discourage you by being overly confrontational. If you are truly interested in giving helpful advice and learning about the language then that is cool.

37. Re: problems with the scoping in eu4

• Posted by jeremy (admin) Jan 09, 2011
• 1758 views

Yes, this works. With a for loop, the variable name is an implicit declaration.

for i = 1 to 5 do 
    ? i 
end for 
? i -- Error... i is not defined. 

Another example

for i = 1 to 3 do 
    ? { 1, i } 
end for 
for i = 4 to 6 do 
    ? { 2, i } 
end for 

Output is:

{ 1, 1 } 
{ 1, 2 } 
{ 1, 3 } 
{ 2, 4 } 
{ 2, 5 } 
{ 2, 6 } 

Jeremy

39. Re: problems with the scoping in eu4

• Posted by jeremy (admin) Jan 09, 2011
• 1758 views

:loop 
  if f = 0 goto end 
  ... 
  goto loop 
  ... 
:end 

Yes. You are correct loops have scope. I was talking more in regards to how a loop jumps and relating that to how with entry also jumps, just like a loop but my statement was not well written, thus incorrect.

Also, as a side note. You do not need all the \\ at the end of your lines. The forum handles making proper paragraphs itself. Simply separate your paragraphs with an empty line.

Jeremy

41. Re: problems with the scoping in eu4

• Posted by mattlewis (admin) Jan 09, 2011
• 1739 views

integer idx = 0  
for idx = 1 to 10 do -- error!  
   ? idx  
end for  

Why do you keep saying this? Why do you believe this? I told you that it wasn't how euphoria works. Have you tried this? It's clear to me you have not. Are you simply trolling? Try this:

procedure main() 
	for i = 1 to 2 do 
		? i 
	end for 
	for i = 3 to 4 do 
		? i 
	end for 
end procedure 
main() 
main() 

Matt

43. Re: problems with the scoping in eu4

• Posted by petelomax Jan 09, 2011
• 1716 views

public function find_all(object needle, sequence haystack, integer start=1) 
    integer kx = 0 
    while start with entry do 
	kx += 1 
	haystack[kx] = start 
	start += 1 
    entry 
	start = find(needle, haystack, start) 
    end while 
 
    haystack = remove( haystack, kx+1, length( haystack ) ) 
 
    return haystack 
end function 

I have to say I'm in the "don't like with entry" crowd. For me the above looks like the while loop will not run for a find_all(,,0) call (I believe bill has the same impression).

I also feel that I might be tempted to code:

public function find_all(object needle, sequence haystack, integer start=1) 
    integer kx = 0 
    if find(needle,ignore_set) then 
        start = 0 
    end if 
    while start with entry do 
        kx += 1 
	haystack[kx] = start 
	start += 1 
    entry 
	start = find(needle, haystack, start) 
    end while 
 
    haystack = remove( haystack, kx+1, length( haystack ) ) 
 
    return haystack 
end function 

It just seems illogical to me that the test occurs first in the source code but is not executed until much later.

Also, can I ask why remove() when haystack=haystack[1..kx] would do the same job, or am I missing something?

The other point that strikes me is that you cannot code:

    while start with entry do 
	integer start 
        ... 
    entry 
	start = find(needle, haystack, start) 
    end while 

Regards, Pete

45. Re: problems with the scoping in eu4

• Posted by petelomax Jan 09, 2011
• 1642 views

You are both inventing this. It is always perfectly ok to replace

  while <condition> with entry do 
    <block a> 
  entry 
    <block b> 
  end while 

with the much clearer

  while true do 
    <block b> 
    if not <condition> then exit end if 
    <block a> 
  end while 

You can like "with entry" and I can dislike it, but please stop saying it removes code duplication or makes things easier to read when it achieves neither.

Regards,
Pete

47. Re: problems with the scoping in eu4

• Posted by jimcbrown (admin) Jan 09, 2011
• 1639 views

Hmm, actually he has a fair point.

This doesn't work

while start with entry do 
	integer start 
	... 
entry 
	start = ... 
end while 

But this does work

while 1 do 
	integer start 
	... 
	if not start then exit end if 
	start = ... 

Hmm.

  while <condition> with entry do 
    <block a> 
  entry 
    <block b> 
  end while 

  while true do 
    <block b> 
    if not <condition> then exit end if 
    <block a> 
  end while 

49. Re: problems with the scoping in eu4

• Posted by SDPringle Jan 10, 2011
• 1567 views

Hello Bill. I am a developer of EUPHORIA and I have tested this software hundreds of times. From experience, some interpreter bugs are difficult to produce. They can be unlikely to show their symptoms in similar contexts. A bug might appear to be fixed on one platform yet be broken on yours. At times the other developers can be skeptical of what you say. However, many of the code samples you supplied contained syntax bugs and it is impossible to track down a bug we cannot reproduce ourselves.

It would be helpful if you included a full source listing with the for and the contents of the ex.err file that gets created. Let us know which version of EUPHORIA you are using. There are several versions 4.0: released, release candidates, betas and alphas.

This source file executes without an error.

include std/console.e 
 
for i = 1 to 10 do  
  puts(1,sprintf("%d, ",2*i) ) 
end for  
for i = 1 to 20 do 
  puts(1,sprintf("%d, ",2*i) ) 
end for  
 
any_key() 

Shawn Pringle