Re: [myhdl-list] attribute enum_encoding: string;

[David H. <da...@ad...>]

> <snip>
> > But "ghdl -a ..." still complains unless I patch m_think.vhd to add "attribute enum_encoding: string;"
> 
> I don't recall all the VHDL rules here off the top
> of my head.  I can test with some synthesis tools
> as well and do some research.  It is a small change
> that can probably be added, but I have to do some
> inquiring before I create a pull-request.

If I could help with testing, I'd love to try any patch to
apply this small change in my branch.  (I had been using
0.8dev from hg before the 0.8 release, to get access to the
fabulous modbv...) I'm just not sure where the best place
is to add the attribute, since it only needs to be added
once and not for every state machine.

> Best I can tell adding
> 
>      attribute enum_encoding: string
> 
> should be harmless, just an additional attribute
> http://quartushelp.altera.com/11.1/mergedProjects/hdl/vhdl/vhdl_file_dir_enum_encoding.htm
> 
> Regards,
> Chris

My motive for adding the attribute is that I found ghdl
will not complain of other potential syntax errors until
after I resolve this one.  In addition to behavioral test
bench code, I apply syntax checks from ghdl, cver,
iverilog, and verilator, because Xilinx's compiler is so
painfully slow...  :)

I've found that if my model passes all those compilers, my
behavioral tests, and Xilinx timing analysis, then my PCI
Express DMA engine has a great chance of working correctly
on the first try when I load it on a dev kit in Linux.  :)