Re: Note on Rev

I'm glad the IBIS and IBIS-X committes are looking at backward
compatibility.

However, there is a required keyword in the IBIS spec called [IBIS ver].
Seems to me it should be pretty straightforward for a parser to treat
old files as requiring the column 1 limitation, while allowing and [End
Notes] keyword at the end of a [Notes] section in future versions.

Thanks for bringing this up, Scott. This limitation has bugged me for a
while, as it has anyone with a programming background (which should be
the majority of us!)

Alan Hilton-Nickel
Transmeta Corp.

Al Davis wrote:
>
> On Friday 06 July 2001 09:55 am, Scott McMorrow wrote:
> ? Stephen,
> ?
> ? I have the following suggestions to eliminate this problem:
> ?
> ? 1) Require that keywords do not appear in any text blocks. Where a
> ? keyword is defined to include the [keyword] syntax. This would be
> ? an unusual circumstance anyway. It is highly unlikely that in
> ? production models that a [keyword] would appear in a text block
> ? sections. These sorts of things generally only occur in examples
> ? and can be easily removed.
> ?
> ? 2) Require an [End Block] keyword for all sections with text
> ? blocks.
> ?
> ?
> ? The requirement that keywords begin in column 1 seems more onerous
> ? to me than either of these suggestions. I would "really" like to
> ? be able to use indented structure within all future models to
> ? facilitate better understandability. Also, I would really like to
> ? not have to deal with a stupid parser that spits out my files when
> ? somehow I managed to create a model that does not position a
> ? keyword in column 1.
>
> We discussed this, both in connector and IBIS-X. The problem is
> backward compatibility.
>
> Under suggestion #1, this basically prohibits any use of "[" in a
> text block. This was the original intended way of doing it. In
> testing, I found that there were too many old files being rejected.
>
> Suggestion #2 is the obviously superior method, except that there are
> lots of files that don't have the end marker, and we want to maintain
> 100% backward compatibility.
>
> Still, this does not require that EVERY keyword begin in column 1.
> The restriction could only apply to those terminating a text block,
> or the text block could be terminated by a line containing a "[" as
> the first non-whitespace character, following a blank line.
 
Received on Fri Jul 20 10:25:56 2001