| Add Proposal | Add Analysis | Edit Class, Environment, or Release |
| Number | 37
|
| Category | errata
|
| Synopsis | Possible syntax restrictions on periods
|
| State | open
|
| Class | errata-discuss
|
| Arrival-Date | Sep 28 2001
|
| Originator | Daryl Stewart
|
| Release | 2001b
|
| Environment |
|
| Description |
I've noticed a few categories in the syntax which use periods which I always assumed had no whitespace near them. Specifically I believe the following nonterminals all contain periods (.) as part of their syntax and require restrictions on preceding and/or following whitespace: The period (.) in design_statement, inst_name, cell_clause and use_clause shall not be preceded or followed by white_space. The period (.) in port, named_parameter_assignment and named_port_connection shall not be followed by white_space. Although I'm not totally sure ;) |
| Fix |
No change. |
| Audit-Trail |
From: Shalom Bresticker <shalom@msil.sps.mot.com> To: Daryl.Stewart@cl.cam.ac.uk Cc: btf-bugs@boyd.com Subject: Re: enhancement/37: Possible syntax restrictions on periods Date: Sun, 30 Sep 2001 09:00:51 +0200 (IST) Daryl, Of the cases you mention, only "port" and "named_port_connection" exist in 1364-1995. I tested them in Verilog-XL, and Verilog-XL allows white space. I also made sure there is white space in those cases in my HTML version of the new BNF. I think the rule is that white space is allowed unless specifically prohibited. The same question also exists for the ":" in use_clause. -- ************************************************************************** Shalom Bresticker Shalom.Bresticker@motorola.com Motorola Semiconductor Israel, Ltd. Tel #: +972 9 9522268 P.O.B. 2208, Herzlia 46120, ISRAEL Fax #: +972 9 9522890 ************************************************************************** On Fri, 28 Sep 2001 Daryl.Stewart@cl.cam.ac.uk wrote: > I've noticed a few categories in the syntax which use periods which I always assumed had no whitespace near them. Specifically I believe the following nonterminals all contain periods (.) as part of their syntax > and require restrictions on preceding and/or following whitespace: > > The period (.) in design_statement, inst_name, cell_clause and use_clause > shall not be preceded or followed by white_space. > The period (.) in port, named_parameter_assignment and named_port_connection > shall not be followed by white_space. > > Although I'm not totally sure ;) From: Shalom Bresticker <Shalom.Bresticker@motorola.com> To: Shalom Bresticker <shalom@msil.sps.mot.com> Cc: Daryl.Stewart@cl.cam.ac.uk, btf-bugs@boyd.com Subject: Re: enhancement/37: Possible syntax restrictions on periods Date: Sun, 30 Sep 2001 10:40:36 +0200 --------------5873AFD3F1D0AE6C534E994B Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Resend. Shalom Bresticker wrote: > Daryl, > > Of the cases you mention, only "port" and "named_port_connection" exist in > 1364-1995. I tested them in Verilog-XL, and Verilog-XL allows white space. > > I also made sure there is white space in those cases in my HTML version of the > new BNF. > > I think the rule is that white space is allowed unless specifically prohibited. > > The same question also exists for the ":" in use_clause. > > -- > ************************************************************************** > Shalom Bresticker Shalom.Bresticker@motorola.com > Motorola Semiconductor Israel, Ltd. Tel #: +972 9 9522268 > P.O.B. 2208, Herzlia 46120, ISRAEL Fax #: +972 9 9522890 > ************************************************************************** > > On Fri, 28 Sep 2001 Daryl.Stewart@cl.cam.ac.uk wrote: > > > I've noticed a few categories in the syntax which use periods which I always assumed had no whitespace near them. Specifically I believe the following nonterminals all contain periods (.) as part of their syntax > > and require restrictions on preceding and/or following whitespace: > > > > The period (.) in design_statement, inst_name, cell_clause and use_clause > > shall not be preceded or followed by white_space. > > The period (.) in port, named_parameter_assignment and named_port_connection > > shall not be followed by white_space. > > > > Although I'm not totally sure ;) -- ************************************************************************** Shalom Bresticker Shalom.Bresticker@motorola.com Motorola Semiconductor Israel, Ltd. Tel #: +972 9 9522268 P.O.B. 2208, Herzlia 46120, ISRAEL Fax #: +972 9 9522890 ************************************************************************** --------------5873AFD3F1D0AE6C534E994B Content-Type: text/html; charset=us-ascii Content-Transfer-Encoding: 7bit <!doctype html public "-//w3c//dtd html 4.0 transitional//en"> Resend. Shalom Bresticker wrote:
Of the cases you mention, only "port" and "named_port_connection" exist
I also made sure there is white space in those cases in my HTML version
I think the rule is that white space is allowed unless specifically
The same question also exists for the ":" in use_clause.
--
On Fri, 28 Sep 2001 Daryl.Stewart@cl.cam.ac.uk wrote:
> I've noticed a few categories in the syntax which use periods which
-- ************************************************************************** Shalom Bresticker Shalom.Bresticker@motorola.com Motorola Semiconductor Israel, Ltd. Tel #: +972 9 9522268 P.O.B. 2208, Herzlia 46120, ISRAEL Fax #: +972 9 9522890 ************************************************************************** --------------5873AFD3F1D0AE6C534E994B-- From: Shalom.Bresticker@motorola.com To: etf-bugs@boyd.com Cc: Subject: Re: errata/37: Possible syntax restrictions on periods Date: Thu, 2 Oct 2003 23:04:22 +0300 (IDT) As previously discussed, in Verilog-XL's implementation of Verilog-1995, spaces ARE allowed in "port" and "named_port_connection". In fact, the only place in the LRM where it states that white space is not allowed (outside of VCD files and $readmem data files) is around the periods separating hierarchy levels in hierarchical name references. And it is not clear why that restriction was added in 2001, when it was not there in 1995. No explanations appear in the archives. In fact, just the opposite. Since then, several people have proposed removing that restriction. (That is issue #257) And in fact, Verilog-XL, VCS, and NCV do not enforce that restriction. In general, the only reason to forbid or require white space is where the lexer would otherwise be confused. That means, within lexical tokens. The question is, what is a lexical token? Section 2 attempts to define lexical tokens, but its description is not complete. The problem comes up with respect to special characters and their combinations. Basically, the only such tokens described in Section 2 are called 'operators', but there are many more such tokens. For example, if I write "a[3:0]", each character there is a separate token, including each of the square brackets and the colon. Another example is simply the semicolon at the end of a statement. It is important to know what character combinations are lexical tokens, because white space is not allowed within lexical tokens (except strings and comments), nor can `define text macros be used for only part of a token. For example, is "+:" in indexed part-selects one token or two? When Paul Graham brought this up once, we all agreed that it is one token, but that is not defined anywhere. I would like to suggest the following: - We add a statement to 2.2 ("White space") that white space is permitted between any tokens, unless specifically forbidden. The first paragraph of 2.1 is probably intended to mean that, but I would like to be more explicit. - We also add a statement to 2.2 that white space is forbidden within tokens (except strings and comments) unless specifically permitted. - Is there a good way to define when white space is required as a token separator? - We should probably replace the mention of operators as a type of lexical token with something more general. - We remove the restriction on white space in hierarchical name references. - We add some sort of wording that tokens are defined by the BNF, in that if characters appear together in the BNF, like "+:", it means that they together form a token. If they have space between them in the BNF, it means that they are separate tokens. - We not restrict use of white space unless necessary. That means, for example, that in library and configuration descriptions, spaces could appear in "-incdir" or in ":config". This might be more controversial. But we might have different rules for library and configuration descriptions, which are not really mainstream BNF. - By the way, 2.8 ("Attributes") is sort of out of place, because attributes are not a lexical token. Shalom On Fri, 28 Sep 2001 Daryl.Stewart@cl.cam.ac.uk wrote: > I've noticed a few categories in the syntax which use periods which I > always assumed had no whitespace near them. Specifically I believe the > following nonterminals all contain periods (.) as part of their syntax > and require restrictions on preceding and/or following whitespace: > > The period (.) in design_statement, inst_name, cell_clause and use_clause > shall not be preceded or followed by white_space. > The period (.) in port, named_parameter_assignment and > named_port_connection shall not be followed by white_space. > > Although I'm not totally sure ;) -- Shalom Bresticker Shalom.Bresticker@motorola.com Design & Reuse Methodology Tel: +972 9 9522268 Motorola Semiconductor Israel, Ltd. Fax: +972 9 9522890 POB 2208, Herzlia 46120, ISRAEL Cell: +972 50 441478 |
| Unformatted |
|
Hosted by Boyd Technology