Re: [vhdl-200x] Standardize GHW format?

On 3/11/2011, at 11:04 AM, David Koontz wrote:

>
> On 2/11/2011, at 10:11 PM, Daniel Kho wrote:
>
>> Hi Martin,
>> So far what I have is just some Ada source codes. There's no documentation on the format. But at least Ada is close enough to VHDL that we could all possibly understand.
>>
>> Yes, I'm having that view as well. It's probably easier to standardize something open source than if it were proprietary / closed source.
>>
>> Perhaps we could get the GHDL and GtkWave authors involved in the standardization effort as well?
>
>
> The source tree for ghdl also contains the C source used in gtkwave. I believe the author of ghdl Tristan Gingold supplied the code to gtkwave's author originally. Three files with ghw in the name in the translate/grt subtree of the source.
>
> As indicated in the second link your first post provided the reason he implemented a new dump file format was because VHDL types weren't conveyable in VCD as it's defined today, being Verilog centric. I'd suggest a little more foundation on the need for dump file format standardization in VHDL before addressing the specifics of a format.
>
> You might find various tools for doing VCD dump file comparison used primarily for regression testing but potentially also usable for comparing different levels of model abstraction. I did a quick search and found no indication of any tools for doing ghw dump file comparison, and we tend to see assertions and textio used in test benches for verification.
>
> The idea being what purpose a dump file standardization serves. I tend to think dump file formats are closely related to simulator implementations and that declaring a standard format that should be adhered to might constitute a burden on existing VHDL simulator implementations. Vendors likely already have their own solutions.
>
> When considering the use of the ghw dump file format I've heard to three waveform viewer implementations, although only gtkwave is actively supported. Is there overriding reason to use this format over another? We see tools capable of performing dump file format conversion, allowing the use of a wider range of formats with waveform viewers.
>
> I can't imagine Verilog and System Verilog implementations embracing changes to VCD to better support VHDL unless they proved transparent. To support VHDL would imply any waveform viewers be modified and they already tend to accept multiple formats (as does gtkwave). The effort to standardize across hardware description languages doesn't seem warranted at first glance.
>
> All this isn't meant as discouragement, rather getting ducks all in a row is in order - proposal, justification, implications. An actual dump file format sounds like only a small part and not exactly rocket science, although there are implications for headers making a format useful for verification.
>
> I forwarded the top of this thread on the eda.org email reflector to Tristan to see if he cares to respond.

From Tristan-

I saw the thread. The copyright isn't a problem. But I don't really understand the need of standardizing a wave format.

Tristan.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.