1. sequence length allocation
I run the following simple program:
sequence s
s =3D "hell"
s[5] =3D 'o'
Most of you will immediately recognize that there is an error here, becau=
se
I allegedly assigned a fifth value to a length 4 sequence. Well, I plead=
guilty, but I appeal because this should not be illegal. Sure, I can use=
append, but the above syntax seems a lot more straightforward. What woul=
d
we loose? Also, I still have a initialize every sequence and specify at
least an initial length. This is not complete dynamic storage allocation=
,
to me.
At least, I feel there should be a sequence initialized as an indefinite
length.
In Euphoria, I still have to initialize the sequence, and tell Euphoria
what length it should be initially. There are tools to append and prepen=
d,
but that only masks the truth that we still have to tell Euphoria what th=
e
length of the sequence is. It seems that if I have a database, with one
sequence with each sub-sequence being a record, I could just simply do
something like this:
s[length(s)+1] =3D new record
I feel this is a more direct approach, than append could ever be. =
Although, it is possible that C, which Euphoria is written in, forces
Euphoria to be like this.
Just a thought.
Alan
=
2. Re: sequence length allocation
Alan wrote:
> sequence s
> s = "hell"
> s[5] = 'o'
Hm. Even I hesistate at that one. 
I'd issue want the program to warning, and add nothing to the sequence.
After all, should:
sequence s
s = 'foo'
s[6] = 'o'
append as well, and create filler for s[4] and s[5]? Should the be blanks,
or zeros?
-- David Cuny
3. Re: sequence length allocation
- Posted by Alan Tu <ATU5713 at COMPUSERVE.COM>
Aug 05, 1998
-
Last edited Aug 06, 1998
>append as well, and create filler for s[4] and s[5]? Should the be
>blanks,
>or zeros?
All right, how about an indefinite sequence. It just seems that I have t=
o
specify the length of the sequence, initialize it. Euphoria has routines=
to make this a ton less painful to manipulate sequences, but still. Your=
example is noted, though. I did consider the possibility of runaway
sequences.
--Alan
=
4. Re: sequence length allocation
-----Original Message-----
De: Alan Tu <ATU5713 at COMPUSERVE.COM>
Para: EUPHORIA at cwisserver1.mcs.muohio.edu
<EUPHORIA at cwisserver1.mcs.muohio.edu>
Fecha: Miércoles 5 de Agosto de 1998 10:32 AM
Asunto: sequence length allocation
>s[length(s)+1] = new record
>I feel this is a more direct approach, than append could ever be.
>Although, it is possible that C, which Euphoria is written in, forces
>Euphoria to be like this.
You can achieve the same with the '&' operator:
s = s & new_record
Perhaps there is a chance to have a &= operator in Euphoria (as
well as +=, -=, /=, etc...)
Regards,
Daniel Berstein
daber at pair.com
5. Re: sequence length allocation
- Posted by Hawke <mdeland at NWINFO.NET>
Aug 17, 1998
-
Last edited Aug 18, 1998
Alan Tu wrote:
> I run the following simple program:
> sequence s
> s = "hell"
> s[5] = 'o'
>
> Most of you will immediately recognize that there
> is an error here, because I allegedly assigned a
> fifth value to a length 4 sequence. Well, I plead
> guilty, but I appeal because this should not be illegal.
I agree here. One caveat tho, it should only be allowed
for 1 index past length(s).
> Also, I still have a initialize every sequence and specify at
> least an initial length. This is not complete dynamic storage
> allocation, to me.
On this note, I disagree. Assigning s={} does not necessarily
give "an initial length". Nor do you even need to do that.
ex:
sequence name
puts(SCR,"what is your name?")
name = gets(KB)
if name[1] = GET_SUCCESS then
name = name[2]
end if
In this example, name is neither initialized NOR given
an initial length... looks pretty dynamic to me... shrug
> At least, I feel there should be a sequence initialized
> as an indefinite length. In Euphoria, I still have to
> initialize the sequence, and tell Euphoria what length
> it should be initially.
errrrrr.... see above?
> It seems that if I have a database, with one
> sequence with each sub-sequence being a record,
> I could just simply do something like this:
> s[length(s)+1] = new record
> I feel this is a more direct approach than append
> could ever be.
Now here I agree. With of course, the caveat mentioned
above. Just like you can slice S[3..2] and come up
{}, you should be able to do the above. Certain routines
could be heavily optimized, simplified and easier to read.
tnx for listening, --Hawke'
6. Re: sequence length allocation
The problem I have with this is that your intention is not known.
If you code...
s[5] = 'o'
the compiler has to guess between two options,
a) You want to extent the sequence,
b) You made a mistake with the index number.
As b) is a real possibility, the safer way of doing
this is to tell the compiler what it is you are trying
to do. Maybe something like...
s[#5] = 'o'
to explicilty and simply tell the compiler that
you are extending the sequence and element 5 is
to be set to 'o'. Of course, what happens to the
elements between the existing last one and the 5th
one is a mystery. So maybe even this ...
s[#] = 'o'
could be used to say "append to the sequence". This
way you wouldn't have to know its current length.
cheers,
Derek Parnell
dparnell @ vic.bigpond.net.au
Melbourne, Australia
-----Original Message-----
From: Alan Tu <ATU5713 at COMPUSERVE.COM>
To: EUPHORIA at cwisserver1.mcs.muohio.edu
<EUPHORIA at cwisserver1.mcs.muohio.edu>
Date: Wednesday, August 5 1998 23:32
Subject: sequence length allocation
|I run the following simple program:
|
|sequence s
|s = "hell"
|s[5] = 'o'
|
|Most of you will immediately recognize that there is an error here, because
|I allegedly assigned a fifth value to a length 4 sequence. Well, I plead
|guilty, but I appeal because this should not be illegal. Sure, I can use
|append, but the above syntax seems a lot more straightforward. What would
|we loose? Also, I still have a initialize every sequence and specify at
|least an initial length. This is not complete dynamic storage allocation,
|to me.
|
|At least, I feel there should be a sequence initialized as an indefinite
|length.
|
|In Euphoria, I still have to initialize the sequence, and tell Euphoria
|what length it should be initially. There are tools to append and prepend,
|but that only masks the truth that we still have to tell Euphoria what the
|length of the sequence is. It seems that if I have a database, with one
|sequence with each sub-sequence being a record, I could just simply do
|something like this:
|
|s[length(s)+1] = new record
|
|I feel this is a more direct approach, than append could ever be.
|Although, it is possible that C, which Euphoria is written in, forces
|Euphoria to be like this.
|
|Just a thought.
|
|Alan
|
