Re: [vhdl-200x] Bugzilla 289: Context declaration requirements are not uniform

Hi Peter,
Yes, that's it.

That just makes me feel the LRM really meant "the selected name of a use
clause" rather than the "logical name of a library clause".

Thanks for pointing that out.

regards, daniel


On 7 November 2013 23:54, Peter Flake <flake@elda.demon.co.uk> wrote:

> Hi Daniel,
>
>
>
> In section 13.2
>
> If two or more logical names having the same identifier (see 15.4) appear
> in library clauses in the same context clause, the second and subsequent
> occurrences of the logical name have no effect.
>
>
>
> Regards,
>
> Peter.
>
>
>
> *From:* owner-vhdl-200x@eda.org [mailto:owner-vhdl-200x@eda.org] *On
> Behalf Of *Daniel Kho
> *Sent:* 02 November 2013 14:04
> *To:* vhdl-200x@eda.org
> *Subject:* Re: [vhdl-200x] Bugzilla 289: Context declaration requirements
> are not uniform
>
>
>
> Typo corrected.
>
>
>
> On 2 November 2013 22:00, Daniel Kho <daniel.kho@gmail.com> wrote:
>
> Hi all,
>
> In Section 13.2, the LRM states that:
> "A given implementation shall provide some mechanism to associate a
> library logical name with a host-dependent library. Such a mechanism is not
> defined by the language."
>
> I think this means that the same logical name should map to a unique
> host-dependent physical library, even if there are multiple repeated
> mentions of the same library clause. This leaves the design library unique
> for the same library logical name.
>
> I.e., the following _library_ clauses:
> lib lib1;
>
> lib lib1;
>
> should point to the same physical design library "lib1".
>
>
>
> Is there anywhere in the LRM that restricts such repeated library clauses?
>
>
>
> regards, daniel
>
>
>
>
>
> On 2 November 2013 17:52, Daniel Kho <daniel.kho@gmail.com> wrote:
>
> Resending, as I believe the previous message wasn't registered.
>
>
>
>
>
> ---------- Forwarded message ----------
> From: *Daniel Kho* <daniel.kho@tauhop.com>
> Date: 2 November 2013 02:52
> Subject: [vhdl-200x] Bugzilla 289: Context declaration requirements are
> not uniform
> To: "vhdl-200x@eda.org" <vhdl-200x@eda.org>
>
>  Hi all,
>
> The submitter points us to Section 13.4 of the LRM:
> "It is an error if, during analysis of a design unit, there is a library
> clause in the expanded context clause of the design unit that occurs as
> part of a replacement of a context reference, and a logical name in that
> library clause denotes a different design library from the design library
> denoted by the logical name during analysis of the context declaration from
> which the library clause was expanded."
>
> After reading this over and over a few times, I too have a few questions.
>
> 1) What is meant by "...from which the library clause was expanded."?
>
> I'm guessing this may be a typo, and what was meant was indeed: "...from
> which the context clause was expanded."
>
> Or am I missing something?
>
>
>
> 2) Another thing that isn't very clear to me is this clause: "if...there
> is a library clause in the expanded context clause of the design unit that
> occurs as part of a replacement of a context reference..."
>
> My view is that a library clause in the expanded context clause couldn't
> possibly _replace_ a context reference. Even if the library clause
> "replaces" part of the expanded context clause (taken from the context
> declaration denoted by the selected name of the context reference in
> question), it should not have any impact on the analysis of the design unit.
>
>
> The same section of the LRM also mentions:
> "NOTE 1—The rules given for use clauses are such that the same effect is
> obtained whether the name of a library unit is mentioned once or more than
> once by the applicable use clauses, or even within a given use clause."
>
> I'm not sure if a similar rule applies to library clauses, but this seems
> to imply so. If for example, the same library logical name is mentioned
> more than once, the effect should be the same as if the library was only
> mentioned once.
>
>
>
> Consider the following example:
>
>
>
> -- contexts c1 and c2 are compiled into the lib1 library:
>
> context c1 is
>
>     library lib1;
>
>     context lib1.c2;
>     ...
>
> end context c1;
>
> context c2 is
>
>     library lib1;
>
>     library lib2;
>     ...
>
> end context c2;
>
>
>
> -- User design:
>
> library lib1;
>
> context lib1.c1;
>
>
>
> The equivalent expanded context clause should look like this (repeated
> library logical names are explicitly written as well):
>
> library lib1;
>
> library lib1;    -- [c1.lib1]: no issues, as this refers to the same
> design library lib1.
>
> library lib1;    -- [c2.lib1]: should also not be an issue? This should
> also refer to the same design library lib1.
>
> library lib2;    -- [c2.lib2]
> ...
>
>
>
> I am guessing that what was really intended by the LRM was to say:
> "if...there is a _use_ clause in the expanded context clause of the design
> unit...".
>
> The reason for my belief comes from my reasoning in 3).
>
>
> 3) The same sentence in the LRM goes on to say: "...and a logical name in
> that library clause denotes a different design library from the design
> library denoted by the logical name during analysis of the context
> declaration..."
>
> I'm not sure how two library clauses that have the same logical name can
> possibly denote different design libraries? The library logical name has to
> be unique, as far as I know.
>
> Perhaps what was meant by the LRM was this:
> "...and a selected name in that use clause denotes a different design
> library from the design library denoted by the selected name during
> analysis of the context declaration..."
>
>
>
> This seems to support my reasoning in 2) that the LRM requires an error to
> be issued when a use clause tries to replace part of the expanded context
> clause for a design unit, AND the selected names for the conflicting use
> clauses are referring to different design libraries.
>
> For example, the following would result in an error (I'm just showing the
> expanded context clause for more clarity):
> library ieee, lib1;
> use lib1.std_logic_1164.std_ulogic;
> context lib1.ctx;    -- assume that the corresponding context declaration
> contains "use ieee.std_logic_1164.std_ulogic"
>
> entity someDesign is port(q:out std_ulogic); end entity someDesign;
>
>
>
> This should result in an error, because although the selected name of the
> use clauses are the same, they refer to different design libraries.
>
> I suggest we re-word the LRM to the following:
> "It is an error if, during analysis of a design unit, there is a use
> clause in the expanded context clause of the design unit that occurs as
> part of a replacement of a context reference, and a selected name in that
> use clause denotes a different design library from the design library
> denoted by the selected name during analysis of the context declaration
> from which the context clause was expanded."
>
>
>
> Or am I missing something else here?
>
>
> The submitter also mentions:
> "Consequently, the same context clause can refer to completely different
> declarations."
>
> I think it is fine for the same context clause to refer to different
> declarations. If the same context clause is used in different designs, I
> believe it is fine for the context clause to refer to different
> declarations, it that is what is intended by the design?
>
> For the submitter's case, if we want Design 2 to have y refer to l1.x.y,
> we would need to specify the library l1 explicitly in the context
> declaration:
>
> context c2 is
>
>     library l1;
>
>     use l1.x.y;
>
> end context c2;
>
>
>
> I think this is not a problem. If we want every design unit to have
> context clauses that always point to the same declarations, then we should
> explicitly specify the library clause within the context declaration.
>
> Otherwise, if we want to allow different designs to use different
> libraries for the same declaration, then we need not specify the library
> clause in the context declaration. Different designs can then use the same
> context reference in their context clauses, but each of those context
> clauses could possibly refer to different design libraries.
>
> I don't see an issue with this, or am I missing something?
>
>
>
> I believe this is pretty similar to the following example:
>
>
>
> -- Say we compiled our non-standard std_logic_1164 package into the
> "my_lib" library:
>
> package std_logic_1164 is
>     type std_ulogic is ('a', 'b', 'c', 'd', true, false);
> end package std_logic_1164;
>
> package body std_logic_1164 is end package body std_logic_1164;
>
> -- This context declaration is also compiled into the "my_lib" library:
>
> context c1 is
>
>     use std_logic_1164.std_ulogic;
>
> end context c1;
>
> -- Design 1:
>
> library ieee;
> use ieee.std_logic_1164;
> context my_lib.c1;    -- equivalent to "use
> ieee.std_logic_1164.std_ulogic;"
>
> entity design1 is port(q:out std_ulogic); end entity design1;
>
>
>
> -- Design 2:
>
> library my_lib;
>
> use my_lib.std_logic_1164;
>
> context my_lib.c1;    -- equivalent to "use
> my_lib.std_logic_1164.std_ulogic;"
>
> entity design2 is port(q:out std_ulogic); end entity design2;
>
>
>
> Both designs use the same type names, but they are actually different
> types that came from different libraries.
>
> This could be intended by a real design as well, where a designer could
> explicitly write the expanded context clauses in place of the context
> clauses that contains context references (as in the comments above).
>
>
>
> Comments?
>
>
> Best regards,
> Daniel
>
>
>
>
>
>
>
>
> --
> This message has been scanned for viruses and
> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
> believed to be clean.
>
> --
> This message has been scanned for viruses and
> dangerous content by *MailScanner* <http://www.mailscanner.info/>, and is
> believed to be clean.
>

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.