On 3/04/11 7:51 PM, Tristan Gingold wrote:
> Thank you for the investigation.  I will try to understand why revision 
> 143 doesn't work.
> 

Thanks for volunteering.  I was estimating around 4 man days to write a map
to find my way around all the code.  I still don't know what all the
contractions mean in names, although I can make educated guesses.

I'll trade you and test my fix to Mr. Landaker's rol/ror bug, submit it as a
bug and a patch.

I'd still like to do an abstract description of ghdl code allowing others to
contribute easier.  As it is today, the knowledge threshold is high.  I
could find the rotate bug easily because the code is reminiscent of what you
would do in VHDL for a CPU design in particular conflating code for
different operators.

> 
> It would be nice to have a nice MacOS-X installer, but I have no experience
>  with that.

I can look into it.  There's nothing wrong with distributing it any format,
but it could use a fairly bulletproof installation script.  It's not an App
and can't really pretend to be without major work building a development
environment.

If you wanted a development environment, I'd suggest the likes of gedit
plugins, gedit and gtkwave being available for all three current platforms
today.

> (and next work could be an IOs version :-)
> 

One little drawback, GPL isn't compatible with App Store developer terms or
end user license.  Couldn't go in the Mac App Store, either.  A third party
would have their copyrights infringed for any contributions to ghdl they
provided under the GPL.

You can of course license as you see fit.  It's a matter of whether or not
you can accept contributions - fixes, enhancements to ghdl and include those
in a closed license distribution.  I'm of the opinion you could attract
contributors under the GPL.  Should you keep contributions to releases
private you'd likely see a fork at some point as long as you continued to
release a GPL version too.

On a more practical note the effort for Apps would be a black hole - all the
engineering on graphics interfaces would be hard to make compatible with
other platforms.  Indeed it's much easier to write an  OS X or IOS
implementation than to make one portable.  There's the Apple Human Interface
Guidelines to consider for IOS and Mac, too.

You'd need to generate your own waveform viewer to make it IOS compatible.
To make the tool effective you might as well add features allow relation
from waveform events to drivers (source code lines).  (See the XMOS Waveform
Viewer demo on Youtube
http://www.youtube.com/watch?v=jyYKjtHjt64&feature=player_embedded#at=73 )
You'd need a source code editing environment and I'd claim you want to add a
control interface to GRT.  The waveform viewer would want to be FIFO
interfaced and you'd want to have the signal dump filter in place.

You could make ghdl 'lite' because of the limitations on an iPad which might
focus on basic VHDL feature compliance.  Some orthogonal issues would
require great attention - more meaningful error messages pointing to source
code.  I think you could garner enough fair use to take single sentences out
of the 2008 LRM semantic restriction descriptions to allow someone to look
up errors there.

(I'll see whether anyone would salute the idea of an appendix for 1076-2013
to standardize error messages as a non-normative reference.  It'd be useful
for tool developers, it would tell them if they are doing enough if nothing
else.  It'll be interesting enough to see IR-1045 dealt with, and this one
goes all the way back to Ada-83 - apostrophe delimiters for attributes
versus aggregates, disambiguating character literals. See Recommended Error
Messages:

http://www.eda.org/twiki/bin/view.cgi/P1076/CollectedRequirements )

The basic message is that today, one ghdl developer may not be enough to get
somewhere with it.

An having gone through tens of ghdl revisions lately, I can appreciate the
work that went into it, and would say:

Thanks!

(And anyone who would like to contribute to the next IEEE 1076 effort can
see http://www.eda.org/vasg/  The header Participation, search the vhdl-200x
reflector for contact info or email Jim Lewis at Synthworks
([email protected]) with your interest.  It's open to individual 
contributors.)







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

Reply via email to