OpenEuphoria

1. Couple of questions about sequences

• Posted by Nevla Aug 12, 2014
• 1187 views

I am a newcomer still going through the Euphoria documentation, learning the basics. I have a couple of questions about sequences:

1) since strings are stored within sequences, and their internal representation consists merely of integers stored within sequences, how can we tell if the content of a sequence is actually a string or merely a series of integers? I understand that from the language's point of view the two are exactly the same thing, but from the programmer's point of view they are not. Is there a way to store a metadata somewhere to indicate that "this is a string" and is meant to be interpreted and used as such?

2) with regard to the use of named constants to reference the indexes of a sequence, I am assuming that there must be some conventional way of doing it, as this appears to be a basic need for handling sequences. From what I understand, the use of named constants for referencing indexes would be indispensable in the case of sequences whose structure is very dynamic - without constantly keeping track of the ever-changing indexes, the contents of a sequence would soon become inaccessible and the whole sequence would become useless. My question is: what's the best way to store, access and update named constants? I guess the end-result would resemble something like a lookup table. But how to implement it? (BTW, I am not asking about maps/associative arrays).

3. Re: Couple of questions about sequences

• Posted by _tom (admin) Aug 12, 2014
• 1179 views

I have just chopped up (literally) the Euphoria documentation and I am trying to put the pieces back together again. I'll try to make a PDF so you can try 4.1 out.

The simple anwer is that there is no string (or character) data-type in Euphoria. There is no metadata in a sequence to turn it into a string data-type.

When integers are small ASCII values there is no way to distinguish a sequence used for numbers or for strings.

The console:display procedure has a heuristic that tries to print text as text, but it can't guess correctly all of the time.

There are UDT (user defined types) for strings and characters in the std/types.e library. Still, you can't definitively sort plain integers from integers used for characters.

Is this a problem? Most of the time it is not. If you put string values into sequence you simply remember to display them as text. This is something you will get used to.

The advantage of Euphoria is that the same sequence operations apply to all sequences making Euphoria easier to use overall.

Indexes are integers; that is it.

There is no dictionary data-type in Euphoria.

Use enums if the sequence is unchanging.

So you do want to use std/map.e if you want dynamically to index an object using a string.

_tom

5. Re: Couple of questions about sequences

• Posted by Nevla Aug 12, 2014
• 1144 views

-- Example 
enum  
    CustID, 
    FirstName, 
    FamilyName, 
    Address, 
    CurrentBalance 
     

Actually I may have used the wrong word by saying "constant". What I meant was the case when named *variables* are used to reference the indexing of some data stored within a sequence whose structure is changing. Suppose the case where your data elements are initially assigned positions 1,2,3,4,5... within a sequence, but later on, because of the changes that take place within the sequence, their indexing position changes. So, in this case I would need to re-assign a new indexing value to each of the named variables, to reflect the new structure. Now, if the "coordinates" of the stored data within a sequence keep changing all the time, and there are several independent elements to keep track of, managing their referencing through named variables may become a challenge. My question was about the best way to maintain an auto-updating sort of lookup table via named variables, to cope with the the changes that take place in the sequence structure.

7. Re: Couple of questions about sequences

• Posted by Nevla Aug 12, 2014
• 1142 views

That I understood. However my question is not being answered: how do I manage efficiently a lookup table that matches named variables with sequence indexes?

9. Re: Couple of questions about sequences

• Posted by _tom (admin) Aug 12, 2014
• 1103 views

Sorry, may reply was terse.

"An index is an integer" means you really are limited to integer expressions to index items in a sequence.

If the sequence is static (meaning no changes in length or shape) then you can make a static list of variables (or constants) with assigned values to index items in a sequence. This is all fixed at the time you write your source-code.

The thing is Euphoria does not allow you to create or destroy variables during the execution of a program. That means you can not have a dynamic list of named variables that index a corresponding dynamic sequence (where the sequence length and shape changes.)

Using the items in a sequence as "variables" used to index another sequence is feasible. The resulting code will likely be messy.

To have the convenience of using some kind of name for indexing means you need a technique that allows you to use an a text string as a key to access a record. Look at the EDS (Euphoria Data Base) or std/map.e for this kind of service. The other example is 3tables.zip (from the RDS Archive) which shows you a simple way to synchronize a sequence of keys with a sequence of records.

Euphoria does not have a syntax trick that lets you use named variables for indexing and then have everything adjust automatically as the sequence changes length and shape.

If this does not answer your question then I need a sample of pseudo-code that demonstrates what you are trying to do.

_tom

11. Re: Couple of questions about sequences

• Posted by Nevla Aug 12, 2014
• 1092 views

Thanks, tom. Your last post perfectly answers my question.