[myhdl-list] Re: VCD generation bug

[Jan D. <ja...@ja...>]

David Brochart wrote:

> 

> Here is the relevant part we get in the VCD file:
> 
> $scope module test $end
> $var real 1 ! a $end
> $scope module dummy_i0 $end
> $var real 1 ! a $end
> $var real 1 " b $end
> $scope module b_i0 $end
> $var real 1 " a $end
> $upscope $end
> $upscope $end
> $upscope $end
> 
> The thing is that we have the same identifier for "a" in "test" and "a" in
> "dummy_i0", and also for "b" in "dummy_i0" and "a" in "b_i0", which is right
> because they are really the same signals. But somehow the version of GTKWAVE I
> run on Linux doesn't like it and doesn't display the waves properly. Which is
> strange because with another version of GTKWAVE on Solaris it works fine.

Right - my gtkwave 1.3.24 on my Linux (RH9)  has the same problem. As the evidence is
that the VCD file is OK, there must be some bug on at least some flavors of Linuxes.

I vaguely remember trying a 2.0 gtkwave version. I'm pretty sure it was OK with
that, so I stopped worrying about the issue and planned to upgrade asap to a
future stable release. Unfortunately, gtkwave 2.0 seems stuck in pre-releases :-(

> Maybe we should give a unique identifier for every signal even if some of the signals
> are actually the same?

Mm, and duplicating all tracing info for every hierarchical level in which the signal is
present? That seems a lot of work and overhead for a workaround for a bug
in another tool. (If required, the solution could be simply to use some older or
newer version for a particular platform.) 

Jan


-- 
Jan Decaluwe - Resources bvba - http://jandecaluwe.com
Losbergenlaan 16, B-3010 Leuven, Belgium
    Python is fun, and now you can design hardware with it:
    http://jandecaluwe.com/Tools/MyHDL/Overview.html