(I am cross-posting this message to the Editorial reflector, as a record for
eventual IBIS 7.x development)
The BCI section of the IBIS 7.0 document has several typographical and
technical errors which will need some clarification and/or correction in a
future IBIS release. Page numbers refer to the Adobe PDF file.
* MM
Technical:
1. The Definition text for BCI_State explains that "The user sets the value
of BCI_State to either "Off" or "Training" on the calls to the Tx and Rx
AMI_Init. The values of BCI_State sent to the Tx and Rx executable models shall
be the same for both the Tx and Rx AMI_Init." I believe that the intended
opening phrase should be "The user, through the EDA tool, sets the value of
BCI_State to either "Off or "Training"..." The EDA tool will be passing the
parameter to the model based on the user's configuration of the EDA tool for
that simulation.
1. Related to this, the BCI_State example shows a List where the first List
option is "Off" and no Default value is specified. Elsewhere in IBIS, the List
format text explains that the first List option is considered the default
unless Default is present. In the definition for BCI_State, the text states
that a model with BCI_State set to "Off" will not execute back channel
adaptation. The example is correct, but some stronger language should be
included to explain that a default of "Off" is recommended as shown. If the
EDA tool is assumed (as explained above) to set BCI_State to "Training" based
on user input, then a default of "Off" will prevent automatic or accidental
entry of the Training state.
1. Related to both the above, the Definition and Usage Rules for BCI_State
are ambiguous as to what should happen if the initial function call to only
*one* of the Tx or Rx models contains BCI_State "Off" when the other model in
the channel has BCI_State set to some other value. Similarly, if only one of
the Tx or Rx models contains BCI_State "Training" on the initial function call,
there's no explicit statement as to what should happen.
Editorial:
1. On page 268, a stray "11" string appears at the end of the Definition
section for BCI_State, as in:
Definition: The user sets the value of BCI_State to either "Off" or "Training"
on the calls to the
Tx and Rx AMI_Init. The values of BCI_State sent to the Tx and Rx executable
models shall be
the same for both the Tx and Rx AMI_Init.11
1. On page 269, an extended double-quote character is used in the example:
Example:
(BCI_Message_Interval_UI(Usage Info) (Type Integer) (Value 2048)
(Description "Training requires at least 2000 UI per adaptation message")
1. On page 269, in the same example, the closing parenthesis character is
missing for BCI_Message_Interval_UI:
Example:
(BCI_Message_Interval_UI(Usage Info) (Type Integer) (Value 2048)
(Description "Training requires at least 2000 UI per adaptation message")
1. On page 270, the same closing parenthesis problem occurs for the
BCI_Training_UI example:
(BCI_Training_UI (Usage In) (Type Integer) (Value 100000)
(Description "BCI training may require 100000 UI")
1. On page 268, there is at least one missing space under Usage Rules, as in:
If theTx AMI_GetWave returns a value in BCI_State...