Hi Paul,

first thank you for taking time to write these test cases.

On Tue, Sep 22, 2009 at 03:56:29PM +0100, Paul Welton wrote:
> Here are three small test cases for the problems encountered in porting
> the processor to GHDL:
> 
> Dimensionless vector constants
[...]
> Results in:
> 
> v1.vhd:35:15: choice must be locally static expression
> v1.vhd:38:15: choice must be locally static expression
> v1.vhd:41:15: choice must be locally static expression
> ghdl: compilation error

Maybe ghdl is overzelous but I think it is right.

> The commented-out alternative code avoids the problem.
> 
> Ambiguous integer operands to mod
[...]
> v2.vhd:13:21: can't resolve overload for operator "mod"
> v2.vhd:13:21: possible interpretations are:
> *std_standard*:1:1: integer
> *std_standard*:1:1: UNIVERSAL_INTEGER
> ghdl: compilation error

Interesting.  I have to check why integer is taken into consideration.

> hread failure for vector of length 8
> ------------------------------------
[...]
> Attempt read ...
> ./e:error: bound check failure at ../../../src/std/textio_body.v93:1311
> ./e:error: simulation failed
> 
> This can be avoided by changing line 8 of the data from "9c" to "9cK".

Even more interesting: I can't reproduce the issue.  Which version of ghdl
were you using ?  (ghdl --version)

Tristan.

_______________________________________________
Ghdl-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/ghdl-discuss

Reply via email to