| Add Proposal | Add Analysis | Edit Class, Environment, or Release |
| Number | 394
|
| Category | errata
|
| Synopsis | 17.10.2: Meaning of special characters in $value$plusargs
|
| State | open
|
| Class | errata-discuss
|
| Arrival-Date | Jul 14 2003
|
| Originator | Gordon Vreugdenhil <gvreugde@Synopsys.COM>
|
| Release | 2001b: 17.10.2
|
| Environment |
|
| Description |
$value$plusargs makes the assumption that leading nulls are ignored in a string variable provided for the match string. The current example in the LRM has (in part): reg [64*8:1] pstring; ... pstring = "TEST%d"; if ($value$plusargs(pstring, testname)) The representation of "pstring" has 58 nulls followed by the 6 characters "TEST%d". There is an assumption that $value$plusargs skips leading nulls in the match string. This should be clarified. In addition, consider other leading characters (whitespace, etc). For example: pstring = " TEST%d"; if ($value$plusargs(pstring, testname)) Is this equivalent to the "TEST%d" for matching purposes? I suggest that we either say something like: 1) The string searched for by $value$plusargs is the sequence of characters which would be found by $sscanf(search_string, "%s", ...). or 2) $value$plusargs ignores leading nulls in the search string provided. The argument in benefit of (2) is that it is necessary to deal with Verilog string assignment semantics but doesn't otherwise change search semantics. It may be reasonable to distinguish between a search for "TEST%d" and " TEST%d"; assumably the command line would have to look like '+ TEST10' to allow the latter, but that is a shell/tool interaction question that we shouldn't worry about. I think that adopting (2) makes the most sense and is minimally restrictive. Gord. -- ---------------------------------------------------------------------- Gord Vreugdenhil gvreugde@synopsys.com Staff Engineer, VCS (Verification Tech. Group) (503) 547-6054 Synopsys Inc., Beaverton OR |
| Fix |
|
| Audit-Trail |
|
| Unformatted |
|
Hosted by Boyd Technology