1. Question on scope
I'm working on outlining a project that is clearly made easier with the
advent of namespacing; however, I do need clarification on the behavior of
include files both with and without namespacing:
If I have
include foo.e
include bar.e
include baz.e
in a Euphoria source file, the public routines of foo.e are _not_ available
to routines in bar.e or baz.e _unless_ those files include foo.e, correct?
Does this still hold if they are included into a namespace?
--
Jeff Zeitlin
jzeitlin at cyburban.com
(ILink: news without the abuse. Ask via email.)
2. Re: Question on scope
On 18 Dec 2001, at 7:12, jzeitlin at cyburban.com wrote:
>
> I'm working on outlining a project that is clearly made easier with the
> advent of namespacing; however, I do need clarification on the behavior of
> include files both with and without namespacing:
>
> If I have
>
> include foo.e
> include bar.e
> include baz.e
>
> in a Euphoria source file, the public routines of foo.e are _not_ available to
> routines in bar.e or baz.e _unless_ those files include foo.e, correct?
All globals in foo are accessable to bar and baz. If you include foo more than
once, or in a nested include, you get an error.
>Does
> this still hold if they are included into a namespace?
I don't know, but it should.
Kat
3. Re: Question on scope
Jeff Zeitlin writes:
> If I have
>
> include foo.e
> include bar.e
> include baz.e
>
> in a Euphoria source file, the public routines of foo.e are _not_ available
> to routines in bar.e or baz.e _unless_ those files include foo.e, correct?
You could make a good case for that behavior, but in fact,
the globals of foo.e are available automatically to bar.e and baz.e,
whether namespaces are defined or not.
Kat writes:
> If you include foo more than
> once, or in a nested include, you get an error.
You can include foo.e again if you want.
All subsequent includes of foo.e will be ignored (no error),
except that you can specify a namespace identifier on
any include statement, and even have multiple identifiers
refering to the same include file.
Regards,
Rob Craig
Rapid Deployment Software
http://www.RapidEuphoria.com
4. Re: Question on scope
On Tue, 18 Dec 2001 12:40:11 -0800, Robert Craig <rds at RapidEuphoria.com>
wrote:
>Jeff Zeitlin writes:
>> If I have
>> include foo.e
>> include bar.e
>> include baz.e
>> in a Euphoria source file, the public routines of foo.e are _not_ available
>> to routines in bar.e or baz.e _unless_ those files include foo.e, correct?
>You could make a good case for that behavior, but in fact,
>the globals of foo.e are available automatically to bar.e and baz.e,
>whether namespaces are defined or not.
This is good to know; it just made part of the project even simpler. I was
prepared to write the project assuming the behavior in my question, but it
would have been more complicated.
>Kat writes:
>> If you include foo more than
>> once, or in a nested include, you get an error.
>You can include foo.e again if you want.
>All subsequent includes of foo.e will be ignored (no error),
>except that you can specify a namespace identifier on
>any include statement, and even have multiple identifiers
>refering to the same include file.
Good to know. Leads to a new question: Suppose I have included foo.e as
bar and as baz. Is it possible to write a procedure that overrides one of
the publics in one of the namespaces? IOW, can I define namespaces other
than by includes? If this is answered in the docs, tell me so - I haven't
had the chance to read them thoroughly, only glanced through them.
--
Jeff Zeitlin
jzeitlin at cyburban.com
(ILink: news without the abuse. Ask via email.)
5. Re: Question on scope
19/12/2001 12:07:25 PM, jzeitlin at cyburban.com wrote:
>
>>You can include foo.e again if you want.
>>All subsequent includes of foo.e will be ignored (no error),
>>except that you can specify a namespace identifier on
>>any include statement, and even have multiple identifiers
>>refering to the same include file.
>
>Good to know. Leads to a new question: Suppose I have included foo.e as
>bar and as baz. Is it possible to write a procedure that overrides one of
>the publics in one of the namespaces? IOW, can I define namespaces other
>than by includes? If this is answered in the docs, tell me so - I haven't
>had the chance to read them thoroughly, only glanced through them.
>
Unfortunately, no.
Namespaces can only be declared via the include statement.
If a file is included multiple times, each with a different namespace
identifier, then all that
happens is that the globals in the include file are only instantiated once but
have multiple
aliases.
Thus is foo.e has a global called "XYZ" then
include foo.e as n1
include foo.e as n2
means that there is only one "XYZ" in memory but it can be referenced as ...
n1:XYZ
or
n2:XYZ
or just
XYZ
if there are no other global XYZ defined in *other* included files or the
current file.
With the example above, the following code ...
n1:XYZ = 0
n2:XYZ = 1
? n1:XYZ
will print out the value 1, because there is one a single XYZ in memory that
both namespaces, n1 and
n2, refer to.
On a personal note, I am very disappointed that RDS chose to implement it this
way. I was looking
forward to having multiple includes instantiate multiple "objects" in memory. Oh
well, may be later.
---------
Derek.
